Liquid biopsy analysis of cell status to predict immunotherapy toxicity

Abstract

A method is disclosed for predicting the likelihood of severe immune-related adverse events (irAEs) occurring in melanoma patients following immunotherapy based on the abundance of activated CD4 memory T cells and / or T cell receptor (TCR) diversity in a peripheral blood sample obtained from the patient. The present invention may include methods and compositions for predicting the likelihood of severe immune-related adverse events (irAEs) occurring in patients undergoing immunotherapy based on biomarkers derived from a peripheral blood sample obtained from the patient prior to receiving immunotherapy.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Application No. 63 / 299,377, filed January 13, 2022, which is incorporated herein by reference in its entirety.

[0002] STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT This invention was made with government support under grants CA187192 and CA238711 awarded by the National Institutes of Health. The government has certain rights in this invention.

[0003] Materials Incorporated by Reference Not applicable.

[0004] FIELD OF THE INVENTION The present disclosure relates generally to methods for predicting immunotherapy toxicity in patients. [Background technology]

[0005] Background of the Invention Although ICIs have revolutionized cancer treatment, severe immune-mediated toxicity commonly occurs in approximately 10% to 60% of melanoma patients treated with ICIs, with the rate of toxicity closely related to the specific therapy administered. ICI-induced toxicity, also known as irAEs, can affect various organ systems, including the lungs, liver, heart, skin, pituitary gland, and gastrointestinal tract, and can be associated with substantial morbidity requiring urgent medical intervention. Such morbidity can lead to the cessation of anticancer treatment and, in the most severe cases, death. The biological drivers of irAEs are not well characterized, and no standard clinical practice exists to identify patients at highest risk for developing irAEs.

[0006] Therefore, several groups have conducted investigations based on blood or tumor analysis for potential biomarkers of ICI-induced toxicity. However, these studies generally focused on early-stage prediction during treatment or on a single organ system, and their predictive performance for irAEs in the pretreatment setting, regardless of the affected organ system, was limited. Recently, a potential irAE biomarker for pneumonitis alone was reported using tumor immunohistochemistry, but this biomarker was indirectly identified by the Cancer Genome Atlas, was not annotated for toxicity, and was evaluated in a case-control setting that did not include low-grade irAEs. Another group identified a single-nucleotide polymorphism in the gene encoding microRNA-146a and associated it with the occurrence of severe irAEs. Furthermore, other groups have identified ICI response biomarkers without irAE investigation.

[0007] Given the considerable heterogeneity of ICI-induced irAEs, including variability in timing, severity, and location, determining the factors that trigger them remains challenging. Pre-existing autoantibodies, autoreactive tissue-resident T cells, and T cells with specificity for viral antigens resulting from chronic viral infections have all been implicated in irAEs. Alterations in the gut microbiome, leading to increased colonic interleukin-1β expression, have also recently been reported in ICI-induced colitis. Given these observations, several groups have investigated the similarities between irAEs and autoimmune diseases. Indeed, case reports have shown that ICIs can induce overt autoimmunity, suggesting that irAEs may be underlying subclinical autoimmunity in a subset of patients. However, it is unclear whether a common immunological condition precedes the distinct manifestations of ICI-induced toxicity. Summary of the Invention [Means for solving the problem]

[0008] Summary of the Invention Various embodiments of the present disclosure include methods and compositions for predicting the likelihood of a patient undergoing immunotherapy experiencing a severe immune-related adverse event (irAE) based on biomarkers derived from a peripheral blood sample obtained from the patient prior to receiving immunotherapy. In one embodiment, the disclosed method includes obtaining a peripheral blood sample from the subject prior to immunotherapy treatment and quantifying the abundance of activated CD4 memory T cells and T cell receptor (TCR) diversity in the peripheral blood sample. A preferred method further includes classifying the patient as likely to experience severe irAR if the combination of the abundance of activated CD4 memory T cells and T cell receptor (TCR) diversity exceeds a threshold value (sometimes referred to herein as a model index). The threshold value can be determined using a model index that identifies the levels of activated CD4 memory T cells and TCR diversity and provides a range of values ​​representing the upper limit above which the patient is susceptible to irAR. In one embodiment, a model index value (combined CD4 memory T cell value and TCR diversity value) above a predetermined threshold predicts more severe irAR. In an additional aspect, the method further comprises determining a threshold value by reference to a known clinical standard. In another aspect, the disclosed method comprises determining the abundance and T cell receptor (TCR) diversity of activated CD4+ memory T cells using at least one of bulk RNA sequencing (CIBERSORTx and MiXCR), mass cytometry by time of flight (CyTOF), immunoSEQ® TCR-β profiling, droplet-based scRNA sequencing and scTCR sequencing, and targeted RNA sequencing using an RNA panel targeting activated CD4+ memory T cells.

[0009] In another aspect of the present disclosure, a method for predicting the likelihood of a severe immune-related adverse event (irAE) occurring in a patient receiving immunotherapy is disclosed. In one aspect, the method includes obtaining a first peripheral blood sample from a subject before receiving immunotherapy treatment and obtaining a second peripheral blood sample after administering immunotherapy. In these other aspects, the disclosed method includes quantifying a first TCR diversity level from the first peripheral blood sample and a second TCR diversity level from the second peripheral blood sample. The disclosed method further includes obtaining a degree of TCR expansion by subtracting the first TCR diversity level from the second TCR diversity level. The disclosed method further includes classifying the patient as likely to develop severe irAR if the degree of TCR expansion exceeds a threshold. In one aspect, the method includes predicting the time point at which severe irAR will occur based on the degree of TCR expansion, where a higher degree of TCR expansion predicts an earlier onset of severe irAR. In one aspect, the method includes determining the level of T cell receptor (TCR) diversity using at least one of bulk RNA sequencing (CIBERSORTx and MiXCR), mass cytometry by time of flight (CyTOF), immunoSEQ® TCR-β profiling, droplet-based scRNA sequencing and scTCR sequencing, and targeted RNA sequencing using an RNA panel targeting activated CD4+ memory T cells.

[0010] Other objects and features will be in part apparent and in part pointed out hereinafter.

[0011] Those skilled in the art will appreciate that the following drawings are for illustrative purposes only and are not intended to limit the scope of the present teachings in any way. [Brief explanation of the drawings]

[0012] [Figure 1]Figure 1 is a schematic overview of the study described in this disclosure, including an overview of the patients included in the study, a summary of their irAE status, exclusion criteria, and downstream analyses performed. Of the 78 total eligible patients, 71 were evaluable for irAE analysis after application of the exclusion criteria. [Figure 2A-B] Figure 2A shows a set of color-coded charts representing the characteristics of the single-cell discovery cohort from Figure 1, including the highest irAE grade and durable clinical response after initiation of immunotherapy. Figure 2B shows a UMAP chart of viSNE projections of peripheral blood cells analyzed by CyTOF. t-SNE, t-distributed stochastic neighbor embedding. [Figure 2C-D] Figure 2C (left) shows a heatmap depicting the relative abundance of 20 cell states identified by CyTOF in 18 patients grouped by future irAE status, and (right) shows a graph depicting the association between cell state abundance and the occurrence of severe irAEs. Statistical significance was determined by a two-tailed, unpaired Wilcoxon rank-sum test and represented by a directional -log10 P value. For associations with no severe irAEs, the -log10 P value was multiplied by -1. Q values ​​were determined by the Benjamini-Hochberg method. Figure 2D shows a graph of CD4 TEM cell (CyTOF) frequency in pretreatment peripheral blood of patients stratified by future irAE status (no severe irAEs, n = 10 patients; severe irAEs, n = 8 patients). The center line, top and bottom edges, and whiskers of the box indicate the median, first and third quartiles, and minimum and maximum values, respectively. Statistical significance was determined by a two-tailed, unpaired Wilcoxon rank sum test. [Figure 3A] Figure 3A is a UMAP of peripheral blood cells profiled by scRNA-seq (Figure 2A) from 13 patients co-analyzed by CyTOF, color-coded by cell type, patient, and condition (n=32). T / NKT, NK-like T cells. [Figure 3B]Figure 3B shows a UMAP of cell state abundance (scRNA-seq) against future irAE status and CD4 T cell frequency (CyTOF). The former was quantified by a two-tailed, unpaired Wilcoxon rank-sum test and expressed as a -log10 P value. For associations with no severe irAEs, the -log10 P value was multiplied by -1. CD4 T cell states 5 and 3 are collectively shown as CD4 T 5+3. [Figure 3C] Figure 3C shows a heatmap of DEGs (Padj<0.05) between CD4 T cell states 5 and 3 and other CD4 T cell states. Within each state, columns represent the mean expression from individual patients converted to z-scores. [Figure 3D] Figure 3D (Left) Frequency of candidate activated and resting subsets of CD4 T5+3 cell status in 13 patients stratified by no severe irAE status (n=7) and severe irAE status (n=6). Patients were considered to express activation markers with counts per million (CPM) >0. Significance was determined by a two-sided, unpaired Wilcoxon rank-sum test, and (Right) receiver operating characteristic curve plots showing the performance of the CD4 T5+3 subset (left panel) to predict severe irAE occurrence. NS, not significant. Two sets of graphs. [Figure 3E] Figure 3E is a graph showing pretreatment TCR clonotype diversity in each T cell status, total T cells, CD8+ T cells, CD4+ T cells, and activated versus resting CD4+ T5+3 cells, grouped by future irAE status (defined in Figure 3D). TCR diversity was calculated for all patients (n=9) with at least 100 TCR clones. Status was ordered by the AUC between TCR diversity and severe irAE status. [Figure 3F]Figure 3F is a color-coded chart of the average expression of key lineage and activation genes across CD4 T cell states. States within the boxes correspond to TEM and TEM-like phenotypes. The center line, top and bottom edges, and whiskers of the box indicate the median, first and third quartiles, and minimum and maximum values, respectively. [Figure 4A-B] Figure 4A shows the association between pretreatment peripheral blood leukocyte composition (CIBERSORTx) and the occurrence of severe irAEs in bulk cohort 1 (n = 26 patients) and bulk cohort 2 (n = 27 patients) (Figure 1). Significance was determined by a two-tailed, unpaired Wilcoxon rank-sum test and expressed as a -log10 P value. For associations with no severe irAEs, the -log10 P value was multiplied by -1. Figure 4B shows the TCR clonotype diversity (Shannon entropy) in both bulk cohorts (n = 53 patients) stratified by future irAE status (no severe irAEs, n = 36; severe irAEs, n = 17). The center line, top and bottom ends of the box, and whiskers indicate the median, first and third quartiles, and minimum and maximum values, respectively. Significance was determined by a two-tailed, unpaired Wilcoxon rank sum test. [Figure 4C] Figure 4C is a graph showing the development of a combined model for predicting severe irAEs, incorporating activated CD4 TM cell abundance and TCR clonotype diversity from pre-treatment peripheral blood transcriptomes, trained on bulk cohort 1 and integrating model scores across both cohorts. High / low score cut-points were optimized using Youden's J statistic for bulk cohort 1. [Figure 4D]Figure 4D shows two sets of graphs: (left) ROC plots showing combined model performance in bulk cohort 2 (holdout validation) regardless of whether it was applied to all patients (both therapies, n = 27), combination therapy patients (n = 11), or PD-1 monotherapy patients (n = 16), and (right) ROC plots showing combined model performance in bulk cohorts 1 and 2 regardless of whether it was trained on PD-1 patients (n = 29) and tested on combination therapy patients (n = 24), or vice versa. The AUC is shown for each ROC curve. [Figure 4E] Figure 4E shows a graph of the combined model score for all bulk cohort patients (n=53) after training the model using LOOCV for severe irAE occurrence (Figure 13), grouped by highest irAE grade per patient. The center line, upper and lower ends of the box, and whiskers indicate the median, first and third quartiles, and minimum and maximum values ​​within a 1.5 × interquartile range of the upper and lower limits of the box, respectively. Statistical significance was determined by Kruskal-Wallis test. [Figure 5A-B] Figure 5A is a graph showing pretreatment predictions of time to onset of severe irAEs in patients treated with combination therapy. Cutpoints were optimized using a composite model score trained with LOOCV. Only patients in bulk cohorts 1 and 2 who did not experience early-stage progression were analyzed (n=23). Statistical significance was assessed by a two-sided log-rank test. Figure 5B is a set of two graphs showing TCR clonal dynamics in patients treated with combination therapy in relation to the occurrence of severe irAEs. Left: Change in TCR clonality from baseline after initiation of combination therapy, measured by 1-Pielou equilibrium. Future irAE status is indicated by color. Right: Same as left, but showing change in clonality according to future irAE status. Significance was determined by a two-sided, unpaired Wilcoxon rank-sum test. [Figure 5C-D]Figure 5C shows the enrichment of the CD4 T 5+3 gene signature in CD4 T cells from pretreatment PBMC samples from the three patients analyzed in Figure 5B. All of these patients developed severe irAEs after ICI initiation and also showed TCR clonal expansion (Figure 5D). The center line, top and bottom edges, and whiskers of the box indicate the median, first and third quartiles, and minimum and maximum values, respectively. Dots represent cells profiled by scRNA-seq and annotated by either Azimuth (CD4 naive, n = 245 cells; CD4 TCM, n = 320 cells) or their clonal persistence from baseline to early time points during treatment (persistent CD4, n = 190 cells). The most persistent CD4 clonotype in this analysis showed evidence of clonal expansion (Figures 15F and G). Significance was determined by a two-tailed, unpaired Wilcoxon rank-sum test for persister cells. ssGSEA, single-sample GSEA. Figure 5D is a graph of the difference in the absence of severe irAEs stratified by the degree of TCR clonal expansion after initiation of combination therapy, as measured by the change in 1-Pielou equilibrium. Patients were grouped into the following tertiles: no clonal expansion (n=5), intermediate (n=5), and high clonal expansion (n=5). Statistical significance was assessed by a two-sided log-rank test. [Figure 6] Figure 6 is a color-coded chart showing a large-scale assessment of circulating leukocytes in autoimmune diseases. Enrichment of circulating leukocyte levels in two autoimmune disorders compared to healthy controls. Leukocyte composition was determined by CIBERSORTx. Significance was determined by a two-tailed, unpaired Wilcoxon rank sum test and pooled meta-z-score. Details of the analysis workflow and underlying dataset are shown in Figure 16. [Figure 7A-C]Figure 7A is a UMAP representation of pretreatment peripheral blood leukocytes from 13 metastatic melanoma patients profiled by droplet-based scRNA-seq (10x Genomics), color-coded by major cell lineage, severe irAE status, TCR expression by scV(D)J-seq, and BCR expression by scV(D)J-seq (related to Figure 3A). Figure 7B is a schematic of unsupervised hierarchical clustering (average linkage) of the mean log2 transcriptome per CD4 T cell cluster identified from scRNA-seq data. Figure 7C is a dot plot showing the mean expression of key activation markers (HLA-DX, MKI67) and lineage markers (SELL, CCR7) in CD4 T cell clusters. [Figure 7D-E] Figure 7D is a graph of unsupervised hierarchical clustering (average linkage) of the mean log2 transcriptome per CD4 T cell cluster identified from scRNA-seq data, showing all pairwise combinations of scRNA-seq clusters within each of the major cell types analyzed (B cells, CD4 T cells, CD8 T cells, NK cells, and monocytes). Across 82 possible pairwise combinations, CD4 T cells achieved the highest Spearman correlation with CD4 TEM levels enumerated by CyTOF and the strongest association with severe irAE occurrence. Cells annotated as "T / NKT" were grouped with CD8 T cells. Figure 7E is a graph of unsupervised hierarchical clustering (average linkage) of the mean log2 transcriptome per CD4 T cell cluster identified from scRNA-seq data, showing all pairwise combinations ranked by the mean of each feature after unit variance normalization (mean 0 and standard deviation 1). In this analysis, the -log10 P values ​​(two-tailed, unpaired Wilcoxon rank sum test) for association with severe irAEs were normalized to unit variance without taking into account the direction of the association. [Figure 8A]Figure 8A shows a UMAP projection of the scRNA-seq data generated in this study embedded labeled with Azimuth using a reference PBMC atlas of 162,000 cells profiled by scRNA-seq and 228 antibodies. [Figure 8B-C] Figure 8B is a confusion matrix showing the concordance between phenotypic labeling determined by marker genes and unsupervised clustering (rows; related to Figures 3A and 7A) and reference-guided annotation using Azimuth (columns). In total, 85% of single cells assigned to major lineage groups by Azimuth (B cells, CD4 T cells, CD8 T cells, NK cells, and monocytes) were assigned to the same identity by reference marker gene scoring. Given the absence of NKT cells in the reference atlas used for Azimuth, the T / NKT cluster defined by the unsupervised analysis was relabeled as CD8 T cells. Figure 8C is a graph showing the same analysis as in Figure 3B but for all 27 phenotypic states identified by Azimuth. Among these states, CD4 TEM was most strongly associated with severe irAEs and CD4 TEM enumerated by CyTOF. The combined CD4 TEM and CD4 proliferative states were also strongly associated with severe irAEs. The latter showed the highest expression of HLA-DX and the lowest expression of SELL (panel d), consistent with an activated CD4 TEM phenotype. [Figure 8D-E] Figure 8D is a dot plot showing key activation and lineage markers among the CD4 T cell states annotated by Azimuth. Figure 8E is a set of violin plots showing protein expression levels by Azimuth using antibody-derived tag (ADT) data, supporting the combination of CD4 TEM and CD4 proliferative states in Figures 8C and F. [Figure 8F]Figure 8F is a grid showing the performance of the top cell subsets identified by Azimuth and unsupervised clustering for predicting severe irAEs. The combined CD4 T 5+3 cluster (Figure 3B) was more significantly associated with severe irAEs and CyTOF than the top reference-derived population (Figure 3C). Statistical significance was calculated using a two-tailed, unpaired Wilcoxon rank-sum test. Data in all panels shown are from the 13 samples profiled by scRNA-seq in Figure 3. [Figure 9A] Figure 9A shows the association between severe irAE occurrence and pretreatment levels of T cell status identified by unsupervised clustering (left) and memory-like T cell status identified by Azimuth (right) in 13 PBMC samples profiled by scRNA-seq (Figures 1 and 3A). Activated cells were defined as cells expressing HLA-DX or MKI67 (CPM > 0); quiescent cells were defined by the absence of HLA-DX and MKI67 expression (CPM = 0). [Figure 9B] Figure 9B is a set of graphs showing an analysis of the association of activated, resting, and parental T cell subsets with severe irAE occurrence. Left: Association between severe irAE occurrence and pretreatment levels of memory T cell subsets, total CD4 and CD8 T cells, and total T cells, as quantified by CyTOF, for all 18 patients analyzed in the single-cell discovery cohort (Figures 1 and 2A). The activated phenotype was defined as CD38+ or HLA-DR+ or Ki67+. The resting phenotype was defined as CD38-HLA-DR-Ki67-. Right: ROC plot showing the performance of activated and resting CD4 TEM subsets (left panel) for predicting severe irAE occurrence. Cell fractions were assessed relative to total PBMC content. Statistical significance in a, b was determined by a two-tailed, unpaired Wilcoxon rank-sum test, and nominal -log10 P values ​​are shown. For associations with no severe irAEs, the −log10 P value was further multiplied by −1. [Figure 10A-B]Figure 10A is a schematic diagram illustrating key TCR diversity metrics and the impact of cell abundance, TCR richness, and distinct clonal repertoires on such metrics. Hypothetical CD4 naive and TEM cell subsets are shown as examples. Triangles indicating size differences are not drawn to scale. Figure 10B plots the average clonality (1-Pielou equilibrium) versus average Shannon entropy for each CD4 T cell state identified by unsupervised clustering of scRNA-seq data. The activated cell-enriched TEM state, CD4 T 5+3 (Figures 3B and C), exhibits increased clonality compared to other CD4 states, as expected for this phenotype, while also exhibiting higher diversity (Shannon entropy), thereby indicating increased richness. [Figure 10C-E] Figure 10C is a schematic diagram showing the distribution of EM-like CD4 T cell states (Figure 3F) using available scTCR clonotype data. Figure 10D is a set of graphs showing the association between severe irAE occurrence and TCR diversity (Shannon entropy) in pseudobulk T cells from pretreatment blood. Shown are all T cell states identified by scRNA-seq (left) and T cell states after removing the EM-like states shown in Figure 10C (no severe irAE, n = 5 patients; severe irAE, n = 4 patients). The top and bottom of the box and the whiskers indicate the median, first and third quartiles, and minimum and maximum values, respectively. Figure 10E is a graph showing the same associations as in Figure 10D, but showing the EM-like states alone. The top and bottom of the box and the whiskers indicate the median, first and third quartiles, and minimum and maximum values, respectively. [Figure 10F-G]Figure 10F shows the area under the curve (AUC) for the association between pretreatment peripheral TCR diversity (Shannon entropy) and the occurrence of severe irAEs. This is shown for all combinations of constituent cell states in e, including the combined CD4 T 5+3 cluster after restricting to activated cells (CPM > 0 for HLA-DX or MKI67). Notably, no other combinations of activated EM-like states achieved an AUC > 0.85 in this analysis. Figure 10G shows the BCR clonotype diversity (Shannon entropy) for each B cell state identified by unsupervised clustering (Figure 3A). In Figures 10B and 10D–F, only patients with at least 100 TCR clones were analyzed (n = 9). For consistency, the same patients were also analyzed in Figure 10G. The top and bottom of the box and the whiskers indicate the median, first and third quartiles, and minimum and maximum values, respectively. [Figure 11A-B] Figure 11A is a graph showing the expression of developmentally regulated marker genes in major CD4 T cell subsets from the LM22 signature matrix (normalized by MAS5). The LM22 reference signature for activated CD4 memory T cells is shown to have a TEM profile. Figure 11B is a graph showing CIBERSORTx versus mass cytometry for enumeration of activated CD4 memory T cells in pretreatment peripheral blood of 17 patients with metastatic melanoma. Linear regression lines with 95% confidence bands are shown. Agreement and significance were determined by Pearson r and two-tailed t-tests, respectively. In this plot, activated CD4 memory T cells quantified by CyTOF were defined by CD38 expression, but other activated CD4 TEM subsets also significantly correlated with CIBERSORTx (Figure 11C). [Figure 11C-D]Figure 11C is a cross-correlation plot of lymphocyte subset frequencies determined by CyTOF and CIBERSORTx. Act., activated. Figure 11D is a cross-correlation plot showing the correlation between activated CD4+ memory T cell levels inferred by CIBERSORTx and 14 memory T cell states, including manually gated CD38+ activated subsets within each population, profiled by CyTOF in PBMCs from 17 patients with metastatic melanoma. [Figure 11E-F] Figure 11E is a scatter plot showing the global correlation of lymphocyte subsets enumerated by CIBERSORTx and flow cytometry in peripheral blood samples from five healthy subjects profiled by bulk RNA-seq. A linear regression line with 95% confidence bands is shown. Agreement and significance were determined by Pearson r and two-tailed t-tests, respectively. Because monocytes were variably underrepresented by cytometry compared with complete blood counts, all results in b–e are expressed relative to total lymphocytes. Figure 11F shows the distribution of activated CD4 memory T cell levels quantified by CyTOF (CD38+, HLA-DR+, or Ki67+ CD4 T cells, n = 28 patients), scRNA-seq (HLA-DX+ or MKI67+ cells within CD4 T clusters 5 and 3, n = 13 patients), and CIBERSORTx (n = 60 patients) across all evaluable samples for irAEs profiled by each modality in this study. The center line, top and bottom edges of the box, and whiskers indicate the median, first and third quartiles, and minimum and maximum values, respectively. Statistical significance was determined by the Kruskal-Wallis test. ns, not significant (P > 0.05). [Figure 12A]Figure 12A shows the association between baseline bulk TCR diversity and the highest observed irAE grade for each patient in bulk cohorts 1 and 2, plotted for Shannon entropy and stratified by treatment type. Patients treated with combination therapy are stratified by future irAE status: no severe irAE (n = 10) vs. severe irAE (n = 14 patients) (left) and irAE grade (right): 0 / 1 (n = 3), 2 (n = 7), 3 (n = 12), and 4 (n = 2). Comparisons between the two groups were assessed by a two-sided, unpaired Wilcoxon rank-sum test. ns, not significant (P > 0.05). Linear regression was applied to evaluate the median of each criterion grouped by irAE grade (inset). The significance of the linear fit was determined by a two-sided t-test. Grades 0 and 1 reflect no toxicity and asymptomatic toxicity, respectively, and were pooled. The line in the middle of the box, the upper and lower limits of the box, and the whiskers indicate the median, the first and third quartiles, and the minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the upper and lower limits of the box, respectively. [Figure 12B]Figure 12B shows the association between baseline bulk TCR diversity and the highest observed irAE grade for each patient in bulk cohorts 1 and 2, plotted for the Gini-Simpson index and stratified by treatment type. Patients treated with the combination therapy are stratified by future irAE status: no severe irAEs (n = 10) vs. severe irAEs (n = 14 patients) (left) and irAE grade (right): 0 / 1 (n = 3), 2 (n = 7), 3 (n = 12), and 4 (n = 2). Comparisons between the two groups were assessed by a two-sided, unpaired Wilcoxon rank-sum test. ns, not significant (P > 0.05). Linear regression was applied to evaluate the median values ​​of each criterion grouped by irAE grade (inset). The significance of the linear fit was determined by a two-sided t-test. Grades 0 and 1 reflect no toxicity and asymptomatic toxicity, respectively, and were pooled. The line in the middle of the box, the upper and lower limits of the box, and the whiskers indicate the median, the first and third quartiles, and the minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the upper and lower limits of the box, respectively. [Figure 12C]Figure 12C shows the association between baseline bulk TCR diversity and the highest observed irAE grade for each patient in bulk cohorts 1 and 2, plotted for Shannon entropy and stratified by treatment type. Patients treated with PD1 monotherapy are stratified by future irAE status: no severe irAEs (n = 26) vs. severe irAEs (n = 3 patients) (left) and irAE grade (right): 0 / 1 (n = 19), 2 (n = 7), 3 (n = 2), and 4 (n = 1). Comparisons between the two groups were assessed by a two-sided, unpaired Wilcoxon rank-sum test. ns, not significant (P > 0.05). Linear regression was applied to evaluate the median of each criterion grouped by irAE grade (inset). The significance of the linear fit was determined by a two-sided t-test. Grades 0 and 1 reflect no toxicity and asymptomatic toxicity, respectively, and were pooled. The line in the middle of the box, the upper and lower limits of the box, and the whiskers indicate the median, the first and third quartiles, and the minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the upper and lower limits of the box, respectively. [Figure 12D] Figure 12D shows the association between baseline bulk TCR diversity and the highest observed irAE grade for each patient in bulk cohorts 1 and 2, plotted for the Gini-Simpson index and stratified by treatment type. Comparisons between the two groups were assessed by a two-sided, unpaired Wilcoxon rank-sum test. ns, not significant (P>0.05). Linear regression was applied to evaluate the median of each criterion grouped by irAE grade (inset). The significance of linear agreement was determined by a two-sided t-test. Grades 0 and 1 reflect no toxicity and asymptomatic toxicity, respectively, and were pooled. The center line, upper and lower ends of the box, and whiskers indicate the median, first and third quartiles, and minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the upper and lower limits of the box, respectively. [Figure 13A-B]Figure 13A is a graph similar to that seen in Figure 4D, but applied to both bulk cohorts (n=53 patients) using leave-one-out cross-validation (LOOCV). Figure 13B is a graph similar to that seen in Figure 4C, but for model scores determined by LOOCV. [Figure 13C] Figure 13C is a plot showing the performance of the combined model and other candidate pre-treatment factors for predicting the occurrence of severe irAEs. As indicated, the combined model was trained on bulk cohort 1 (BC1) and validated on bulk cohort 2 (BC2), or vice versa. [Figure 13D] Figure 13D is a graph showing the performance of the combined model trained on bulk cohort 1 for predicting severe irAEs in different patient subgroups from bulk cohort 2. DCB, durable clinical benefit; NDB, non-durable clinical benefit; GI, gastrointestinal. [Figure 13E-F] Figure 13E shows the composite model scores determined by LOOCV for all bulk cohort patients (n = 24) treated with combination therapy, stratified by future irAE grade: 0 / 1 (n = 3), 2 (n = 7), 3 (n = 12), and 4 (n = 2). The central line, upper and lower ends of the box, and whiskers indicate the median, first and third quartiles, and minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the upper and lower limits of the box, respectively. Statistical significance was determined by Kruskal-Wallis test. Figure 13F shows model performance for predicting the occurrence of grade 2+, 3+, or 4 irAEs in combination therapy patients using the scores in Figure 13E. [Figure 13G-H]Figure 13G shows the composite model scores determined by LOOCV for both bulk cohorts (n = 53 patients) for the number of symptomatic irAEs (grade 2+) per patient. The central line, upper and lower ends of the box, and whiskers indicate the median, first and third quartiles, and the minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the box limits, respectively. Statistical significance was determined by the Kruskal-Wallis test. Figure 13H shows the composite model scores determined by LOOCV for both bulk cohorts (n = 53 patients) for the number of organ system toxicities per patient. The central line, upper and lower ends of the box, and whiskers indicate the median, first and third quartiles, and the minimum and maximum values ​​within 1.5 × IQR (interquartile range) of the box limits, respectively. Statistical significance was determined by the Kruskal-Wallis test. [Figure 13I] Figure 13I is a plot showing the distribution of irAEs across patients and organ systems. Patients in bulk cohorts 1 and 2 are organized by decreasing composite model scores determined by LOOCV. The line separating high / low scores was optimized using LOOCV. [Figure 13J] Figure 13J is a graph showing the fraction of patients in both bulk cohorts who developed irAEs in at least two organ systems and the fraction who did not, stratified by the thresholds in Figure 13I. Significance was determined by two-sided Fisher's exact test. [Figure 14A]Figure 14 is a set of graphs showing composite model performance for predicting time to severe irAEs in validation bulk cohort 2. Figures 14a–c show Kaplan–Meier analyses of the absence of severe irAEs for patients treated with combination or PD1 immune checkpoint blockade (a), combination therapy (b), or PD1 monotherapy (c) in bulk cohort 2, stratified by composite model score. Statistical significance was calculated using a two-sided log-rank test. For all panels, training was performed on bulk cohort 1, and cutpoints for the prediction of severe irAEs were optimized for bulk cohort 1 using Youden's J statistic. Notably, analyses a–c focus on the time between treatment initiation and 3 months after treatment initiation, during which all severe irAEs occurred. Kaplan–Meier plots are shown up to 4 months, accounting for extended follow-up of patients who did not experience any severe irAEs. Figure 14A shows a Kaplan-Meier analysis of the absence of severe irAEs for patients treated with combination or PD1 immune checkpoint blockade in bulk cohort 2, stratified by composite model score. Statistical significance was calculated using a two-sided log-rank test. For all panels, training was performed on bulk cohort 1, and cutpoints for the prediction of severe irAEs were optimized for bulk cohort 1 using Youden's J statistic. Notably, this analysis focused on the period between treatment initiation and 3 months after treatment initiation, during which all severe irAEs occurred. The Kaplan-Meier plot is shown up to 4 months, taking into account the extended follow-up of patients who did not experience any severe irAEs. [Figure 14B]Figure 14B is a graph showing a Kaplan-Meier analysis of the absence of severe irAEs for patients treated with combination therapy in bulk cohort 2, stratified by composite model score. Statistical significance was calculated using a two-sided log-rank test. For all panels, training was performed on bulk cohort 1, and cutpoints for the prediction of severe irAEs were optimized for bulk cohort 1 using Youden's J statistic. Notably, the analysis in Figures 14A-C focuses on the period between treatment initiation and 3 months after treatment initiation, during which all severe irAEs occurred. Kaplan-Meier plots are shown up to 4 months, taking into account extended follow-up of patients who did not experience any severe irAEs. [Figure 14C] Figure 14C shows a Kaplan-Meier analysis of the absence of severe irAEs for patients treated with PD1 monotherapy in bulk cohort 2, stratified by composite model score. Statistical significance was calculated using a two-sided log-rank test. For all panels, training was performed on bulk cohort 1, and cutpoints for the prediction of severe irAEs were optimized for bulk cohort 1 using Youden's J statistic. Notably, analyses in a–c focus on the period between treatment initiation and 3 months after treatment initiation, during which all severe irAEs occurred. Kaplan-Meier plots are shown up to 4 months, taking into account extended follow-up of patients who did not experience any severe irAEs. [Figure 15A-B] Figure 15A is a graph showing the balance (Pielou's index) of TCR repertoires assembled by MiXCR (bulk RNA-seq) and immunoSEQ® (genomic DNA) from paired pre-treatment PBMC samples (n = 15 combination therapy patients). Concordance and significance were determined by Spearman rho and two-tailed t-tests, respectively. Figure 15B is a graph similar to that in Figure 5B, but showing clonality for each pre-treatment and on-treatment PBMC sample. Statistical significance was determined by two-tailed, paired Wilcoxon rank-sum test. ns, not significant (P > 0.05). [Figure 15C] Figure 15C is a graph showing the fraction of pretreatment peripheral blood TCR clonotypes detected during treatment in 15 combination therapy patients stratified by no severe irAE status (n=6) and severe irAE status (n=9). Clonotypes with matching proliferative CDR3 β-strand nucleotide sequences were considered identical. The central line, upper and lower ends of the box, and whiskers indicate the median, first and third quartiles, and minimum and maximum values, respectively. Significance was determined by a two-sided, unpaired Wilcoxon rank-sum test. [Figure 15D-E] Figure 15D is a schematic showing cross-referencing of persister T cell clones identified by immunoSEQ® with scTCR-seq and scRNA-seq data from pre-treatment PBMCs from the same three patients (YUALOE, YUNANCY, and YUHONEY), all of whom received combination therapy and developed severe ICI-induced toxicity. Figure 15E is a dot plot showing the log2 expression of key lineage and activation markers across major T cell states annotated by Azimuth, along with persister clones classified into CD4 and CD8 T cells. [Figure 15F] Figure 15F is a graph showing the aggregate change from baseline in productive frequency of persistent clonotypes stratified by lineage (n=2 cell types) and patient (n=3). The sum of the productive frequency differences (% on treatment - % pre-treatment) was calculated from the immunoSEQ® data. Bars represent the mean + / - SD. [Figure 15G] Figure 15G is a set of graphs showing peripheral blood TCR-β profiling using immunoSEQ®. Top: Changes in bulk TCR clonality from baseline (Figure 5b). Bottom: Same as Figure 15F, but showing underlying clonotypes, with circle size proportional to pre-treatment clonal frequency (immunoSEQ®). [Figure 15H] FIG. 15H is a graph similar to that seen in FIG. 5D, but restricted to blood draws obtained on day 1 of cycle 1 of combination therapy and <1 month later (n=7 patients). [Figure 16] Figure 16 is a schematic of a large-scale assessment of peripheral blood leukocytes in autoimmune disorders relative to healthy controls. A diagram (Figure 6) describes the workflow and statistical meta-analysis for assessing the enrichment of individual circulating leukocyte subsets in autoimmune disorders compared to healthy controls. Briefly, CIBERSORTx was applied to enumerate 15 leukocyte subsets in bulk RNA-seq or microarray profiles of peripheral blood samples from patients with either systemic lupus erythematosus (SLE; n = 239) or inflammatory bowel disease (IBD; n = 348) compared to healthy controls. For each dataset and cell subset, a two-tailed, unpaired Wilcoxon rank-sum test was applied to assess the relative abundance differences between the healthy and disease phenotypes. Results were then combined across tests using a meta-z statistic. [Figure 17A] Figure 17A is a schematic diagram showing the gating hierarchy and staining results for CD4 T cell subsets and NKT cells profiled by CyTOF from pre-treatment PBMCs. All CD4 T cell subsets except regulatory T cells (Tregs) were gated similarly to CD8 T cells. TCM, central memory T cells; TEM, effector memory T cells; EMRA, CD45RA+ terminally differentiated effector memory T cells. [Figure 17B] FIG. 17B is a schematic diagram showing the gating hierarchy and staining results for activated and resting CD4 TEM cells profiled by CyTOF from pre-treatment PBMCs. [Figure 17C] FIG. 17C is a schematic diagram showing the gating hierarchy and staining results for monocyte subsets profiled by CyTOF from pre-treatment PBMCs. [Figure 17D] FIG. 17D is a schematic diagram showing the gating hierarchy and staining results for B cell subsets profiled by CyTOF from pre-treatment PBMCs. [Figure 17E]FIG. 17E is a schematic diagram showing the gating hierarchy and staining results for NK cell subsets profiled by CyTOF from pre-treatment PBMCs. [Figure 18A-B] Figure 18 is a set of graphs showing a comparison of automated and manual cell state quantification from CyTOF data. Figure 18A is a scatter plot showing agreement between automated gating (Astrolabe) and manual gating for the indicated peripheral blood cell types. Agreement was assessed by Pearson correlation and linear regression (95% confidence bands are shown). Statistical significance was assessed using a two-tailed t-test. Data are from patients analyzed by CyTOF in Figure 1 (n=18). Figure 18B is a scatter plot similar to that seen in Figure 18A, but for CD4 TEM cells. A representative gating scheme for CD4 TEM is shown in Figure 7A. Agreement was assessed by Pearson correlation and linear regression (95% confidence bands are shown). Statistical significance was assessed using a two-tailed t-test. Data are from patients analyzed by CyTOF in Figure 1 (n=18). [Figure 18C] Figure 18C is a graph showing the association between pretreatment CD4 T cell abundance, expressed as a fraction of total PBMCs, total T cells, or CD4 T cells, and the occurrence of severe irAEs. The center line, top and bottom edges of the box, and whiskers represent the median, first and third quartiles, and minimum and maximum values, respectively. Data are from patients analyzed by CyTOF in Figure 1 (n=18). [Figure 19] FIG. 19 is a schematic diagram of the method described in this disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0013] Detailed Description of the Invention Severe immune-related adverse events (irAEs) occur in approximately 60% of melanoma patients treated with immune checkpoint inhibitor (ICI) combinations and are a major cause of treatment-related morbidity and mortality. However, no reliable method exists for predicting the occurrence or timing of severe irAEs.

[0014] Pre- and intra-treatment analysis of cellular status and T cell receptors predicts the onset and timing of immunotherapy toxicity. Specifically, the abundance of activated CD4+ effector memory T cells in peripheral blood and the diversity of the T cell receptor repertoire combine to generate a composite biomarker predictive of immunotherapy toxicity. Clonal expansion from pre- to intra-treatment predicts the timing of severe toxicity. Targeted RNA sequencing panels enable this analysis in a practical and cost-effective manner.

[0015] Immunotherapy toxicities (immune-related adverse events) can be severe, dangerous, life-threatening, and fatal. In fact, no method exists to reliably predict these toxicities early or before treatment. This would facilitate toxicity prediction, earlier intervention, and more personalized and precise immunotherapy.

[0016] In various embodiments, a method for predicting immunotherapy toxicity in a patient is disclosed. The disclosed method is based on the discovery that two factors derived from analysis of peripheral blood samples, including the abundance level of activated CD4 memory T cells and bulk TCR diversity, are strongly correlated with the occurrence of severe immune-related adverse events (irAEs). In various embodiments, a liquid biopsy method for predicting immunotherapy toxicity in a patient is disclosed, comprising obtaining a peripheral blood sample from a subject prior to immunotherapy treatment. In various embodiments, the method further comprises quantifying the abundance of activated CD4 memory T cells and T cell receptor (TCR) diversity in the peripheral blood sample. In some embodiments, the method further comprises determining a model index predicting the likelihood of the patient developing severe irAR, the model index comprising a combination of the abundance of activated CD4 memory T cells and the diversity of T cell receptor (TCR). The method further comprises classifying the patient as having a high likelihood of developing severe irAR if the value of the model index exceeds a threshold. In some embodiments, the method further comprises predicting the severity of the irAR based on the value of the model index, with a higher model index value predicting a more severe irAR. The threshold for the higher model index value can be determined empirically or by reference to a known clinical standard.

[0017] In various other aspects, a method for predicting the toxicity of immunotherapy in a patient is disclosed based on the degree of TCR expansion, defined herein as an increase in TCR diversity compared to pre-treatment TCR diversity during the initial stage of immunotherapy. The method includes obtaining a first peripheral blood sample from the patient before the start of immunotherapy and obtaining a second peripheral blood sample during the initial stage of administering immunotherapy to the patient. The method further includes obtaining a first TCR diversity from the first peripheral blood sample and a second TCR diversity from the second peripheral blood sample. The method further includes subtracting the first TCR diversity from the second TCR diversity to obtain a degree of TCR expansion. In some aspects, the method further includes classifying the patient as likely to develop severe irAR if the degree of TCR expansion exceeds a threshold. In some aspects, the method further includes predicting the time point at which severe irAR will occur based on the degree of TCR expansion.

[0018] molecular manipulation The following definitions and methods are presented to better define the present invention and for guidance to those of ordinary skill in the art regarding the practice of the present invention. Unless otherwise specified, terms are to be understood according to conventional usage by those of ordinary skill in the art.

[0019] The terms "heterologous DNA sequence," "exogenous DNA segment," or "heterologous nucleic acid," as used herein, refer to a sequence that originates from a source foreign to a particular host cell, or, if from the same source, has been modified from its original form. Thus, a heterologous gene in a host cell encompasses a gene that is endogenous to a particular host cell but has been modified, for example, by the use of DNA shuffling. The term also encompasses non-naturally occurring multiple copies of a naturally occurring DNA sequence. Thus, the term refers to a DNA segment that is foreign or heterologous to the cell, or a DNA segment that is homologous to the cell but is present in the host cell's nucleic acid at a location where the element is not normally found. The exogenous DNA segment is expressed to produce an exogenous polypeptide. A "homologous" DNA sequence is a DNA sequence that is naturally associated with the host cell into which it is introduced.

[0020] Expression vector, expression construct, plasmid, or recombinant DNA construct is generally understood to refer to a nucleic acid with a set of specific nucleic acid elements that are produced by human intervention, including recombinant means or direct chemical synthesis, and that allow for the transcription or translation of a specific nucleic acid in, for example, a host cell. An expression vector can be part of a plasmid, a virus, or a nucleic acid fragment. Generally, an expression vector can include a nucleic acid to be transcribed that is operably linked to a promoter.

[0021] A "promoter" is generally understood to be a nucleic acid control sequence that directs transcription of a nucleic acid. An inducible promoter is generally understood to be a promoter that mediates transcription of an operably linked gene in response to a specific stimulus. A promoter may include necessary nucleic acid sequences near the transcription start site, for example, a TATA element in the case of a polymerase II type promoter. A promoter may also include distal enhancer or repressor elements as needed, which may be located as far as several thousand base pairs from the transcription start site.

[0022] " Transcribable nucleic acid molecule " as used herein refers to any nucleic acid molecule that can be transcribed into RNA molecule.Methods are known for introducing constructs into cells so that transcribable nucleic acid molecules are transcribed into functional mRNA molecules that are translated and thus expressed as protein products.Constructs can also be constructed to allow the expression of antisense RNA molecules in order to inhibit the translation of specific RNA molecules of interest. Conventional compositions and methods for preparing and using constructs and host cells for the practice of the present disclosure are well known to those of skill in the art (see, e.g., Sambrook and Russell (2006) Condensed Protocols from Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, ISBN-10: 0879697717; Ausubel et al. (2002) Short Protocols in Molecular Biology, 5th ed., Current Protocols, ISBN-10: 0471250929; Sambrook and Russell (2001) Molecular Cloning: A Laboratory Manual, 3d ed., Cold Spring Harbor Laboratory Press, ISBN-10: 0879695773; Elhai, J. and Wolk, CP 1988. Methods in Enzymology 167, 747-754).

[0023] The "transcription start site" or "start site" is the position surrounding the first nucleotide that is part of the transcribed sequence, also defined as the +1 position. All other sequences of the gene and its regulatory regions can be numbered relative to this start site. Downstream sequences (i.e., additional protein-coding sequences in the 3' direction) can be designated with a plus, and upstream sequences (the majority of the regulatory region in the 5' direction) are designated with a minus.

[0024] "Operably linked" or "functionally linked" preferably refers to the association of nucleic acid sequences on a single nucleic acid fragment in which the function of one is affected by the other. For example, a regulatory DNA sequence is said to be "operably linked" or "associated" with a DNA sequence encoding an RNA or polypeptide if the two sequences are positioned such that the regulatory DNA sequence affects the expression of the coding DNA sequence (i.e., the coding sequence or functional RNA is under the transcriptional control of the promoter). The coding sequence may be operably linked to the regulatory sequence in a sense or antisense orientation. The two nucleic acid molecules may be part of a single, contiguous nucleic acid molecule or may be adjacent. For example, a promoter is operably linked to a gene of interest if it regulates or mediates the transcription of the gene of interest in a cell.

[0025] A "construct" is generally understood to be any recombinant nucleic acid molecule, such as a plasmid, cosmid, virus, autonomously replicating nucleic acid molecule, phage, or linear or circular single- or double-stranded DNA or RNA nucleic acid molecule, derived from any source, capable of genomic integration or autonomous replication, and comprising one or more operably linked nucleic acid molecules.

[0026] The constructs of the present disclosure may contain a promoter operably linked to a transcribable nucleic acid molecule operably linked to a 3' transcription termination nucleic acid molecule. The constructs may also include additional regulatory nucleic acid molecules, for example, but not limited to, from the 3' untranslated region (3'UTR). The constructs may also include, but are not limited to, the 5' untranslated region (5'UTR) of an mRNA nucleic acid molecule, which may play an important role in translation initiation and may also be a genetic component within the expression construct. These additional upstream and downstream regulatory nucleic acid molecules may be derived from native or heterologous sources relative to the other elements on the promoter construct.

[0027] The term "transformation" refers to the transfer of a nucleic acid fragment into the genome of a host cell, thereby resulting in genetically stable inheritance. Host cells containing the transformed nucleic acid fragments are referred to as "transgenic" cells, and organisms containing transgenic cells are referred to as "transgenic organisms."

[0028] "Transformed," "transgenic," and "recombinant" refer to a host cell or organism, such as a bacterium, cyanobacterium, animal, or plant, into which a heterologous nucleic acid molecule has been introduced. As is generally known and disclosed in the art, the nucleic acid molecule can be stably integrated into the genome (Sambrook 1989; Innis 1995; Gelfand 1995; Innis & Gelfand 1999). Known PCR methods include, but are not limited to, those using paired primers, nested primers, single-specific primers, degenerate primers, gene-specific primers, vector-specific primers, partially mismatched primers, and the like. The term "untransformed" refers to a normal cell that has not undergone a transformation process.

[0029] "Wild-type" refers to a virus or organism as found in nature, without any known mutations.

[0030] The design, production, and testing of variant nucleotides and their encoded polypeptides that have the desired percent identity to the expressed protein and retain the desired activity of the expressed protein are within the skill of the art.For example, directed evolution and rapid isolation of variants can be performed according to the methods described in references including, but not limited to, Link et al. (2007) Nature Reviews 5(9), 680-688; Sanger et al. (1991) Gene 97(1), 119-123; Ghadessy et al. (2001) Proc Natl Acad Sci USA 98(8) 4552-4557.Therefore, those skilled in the art can generate a large number of nucleotide and / or polypeptide variants that have at least 95-99% identity to the reference sequences described herein, and screen these variants for desired phenotypes according to conventional methods in the art.

[0031] The percent (%) of nucleotide and / or amino acid sequence identity is understood to mean the percentage of nucleotides or amino acid residues that are identical to the nucleotides or amino acid residues in the candidate sequence compared to the reference sequence when the two sequences of the candidate sequence and the reference sequence are aligned.To determine percent identity, the sequences are aligned, and gaps are introduced if necessary to achieve maximum percent sequence identity.The sequence alignment procedure for determining percent identity is well known to those skilled in the art.In many cases, sequences are aligned using publicly available computer software such as BLAST, BLAST2, ALIGN2, or Megalign (DNASTAR) software.Those skilled in the art can determine the appropriate parameters for measuring alignment, including any algorithms required to achieve maximum alignment over the entire length of the sequences being compared. Once the sequences are aligned, the percent sequence identity of a given sequence A to a given sequence B (which can alternatively be expressed as a given sequence A having or containing a certain percent sequence identity to a given sequence B) can be calculated as percent sequence identity = X / Y100, where X is the number of residues scored as identical matches by a sequence alignment program or algorithm aligning A and B, and Y is the total number of residues in B. If the lengths of sequence A and sequence B are not equal, then the percent sequence identity of A to B will not equal the percent sequence identity of B to A.

[0032] Generally, conservative substitutions can be made at any position as long as the desired activity is maintained. So-called conservative substitutions can be made where the replaced amino acid has similar properties to the original amino acid, such as Glu for Asp, Gln for Asn, Val for Ile, Leu for Ile, and Ser for Thr. For example, amino acids with similar properties can be aliphatic amino acids (e.g., glycine, alanine, valine, leucine, isoleucine); hydroxyl or sulfur / selenium-containing amino acids (e.g., serine, cysteine, selenocysteine, threonine, methionine); cyclic amino acids (e.g., proline); aromatic amino acids (e.g., phenylalanine, tyrosine, tryptophan); basic amino acids (e.g., histidine, lysine, arginine); or acidic amino acids and their amides (e.g., aspartic acid, glutamic acid, asparagine, glutamine). Deletion is a direct substitution of an amino acid. The position of deletion includes the end of polypeptide and the junction between individual protein domains.Insertion is the introduction of amino acids into polypeptide chains, where direct bonds are formally replaced by one or more amino acids.Amino acid sequences can be modulated using computer simulation programs known in the art, thereby producing polypeptides with, for example, improved activity or altered regulation.Based on these artificially generated polypeptide sequences, the corresponding nucleic acid molecules encoding such modulated polypeptides can be synthesized in vitro using the specific codon usage of desired host cells.

[0033] "Highly stringent hybridization conditions" are defined as hybridization at 65°C in 6xSSC buffer (i.e., 0.9M sodium chloride and 0.09M sodium citrate). Given these conditions, the melting temperature (T) of a DNA duplex between two sequences determines whether a given set of sequences will hybridize. mThe determination can be made by calculating T. If a particular duplex has a melting temperature below 65°C under salt conditions of 6xSSC, the two sequences will not hybridize. On the other hand, if the melting temperature is above 65°C under the same salt conditions, the sequences will hybridize. In general, the melting temperature of any hybridized DNA:DNA sequence can be calculated using the following formula: T m =81.5℃+16.6(log 10 [Na + ]) + 0.41 (fractional G / C content) - 0.63 (% formamide) - (600 / 1). Furthermore, the T of DNA:DNA hybrids can be determined using m decreases by 1–1.5°C with every 1% decrease in nucleotide identity (see, e.g., Sambrook and Russell, 2006 ).

[0034] Transformation of host cells can be carried out using a variety of standard techniques known in the art (see, for example, Sambrook and Russell (2006) Condensed Protocols from Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, ISBN-10: 0879697717; Ausubel et al. (2002) Short Protocols in Molecular Biology, 5th ed., Current Protocols, ISBN-10: 0471250929; Sambrook and Russell (2001) Molecular Cloning: A Laboratory Manual, 3d ed., Cold Spring Harbor Laboratory Press, ISBN-10: 0879695773; Elhai, J. and Wolk, CP 1988. Methods in Enzymology 167, 747-754). Such techniques include, but are not limited to, viral infection, calcium phosphate transfection, liposome-mediated transfection, biolistic-mediated delivery, receptor-mediated uptake, cell fusion, electroporation, etc. The transfected cells can be selected and expanded to result in recombinant host cells containing the expression vector stably integrated into the genome of the host cells. [Table 1-1] [Table 1-2]

[0035] Exemplary nucleic acids that can be introduced into host cells include, for example, DNA sequences or genes from another species, or even genes or sequences that originate from or exist in the same species but are incorporated into the recipient cell by genetic engineering methods. The term "exogenous" also refers to genes that are not normally present in the cell being transformed, or perhaps simply not present in the form, structure, etc. found in the transforming DNA segment or gene, or genes that are normally present and that are desired to be expressed in a manner different from their natural expression pattern, e.g., highly expressed. Thus, the term "exogenous" gene or DNA refers to any gene or DNA segment that is introduced into a recipient cell, regardless of whether a similar gene may already be present in such a cell. The type of DNA contained in exogenous DNA includes DNA already present in the cell, DNA from another individual of the same type of organism, DNA from a different organism, or DNA produced from an external source, such as a DNA sequence containing an antisense message of a gene or a DNA sequence encoding a synthetic or modified version of a gene.

[0036] Host strains generated according to the techniques described herein can be evaluated by several means known in the art (see, e.g., Studier (2005) Protein Expr Purif. 41(1), 207-234; Gellissen, ed. (2005) Production of Recombinant Proteins: Novel Microbial and Eukaryotic Expression Systems, Wiley-VCH, ISBN-10: 3527310363; Baneyx (2004) Protein Expression Technologies, Taylor & Francis, ISBN-10: 0954523253).

[0037] Methods for downregulating or silencing genes are known in the art. For example, expressed protein activity can be downregulated or eliminated using antisense oligonucleotides, protein aptamers, nucleotide aptamers, and RNA interference (RNAi) (e.g., small interfering RNA (siRNA), short hairpin RNA (shRNA), and microRNA (miRNA) (e.g., hammerhead ribozymes and short hairpin RNAs are described in Fanning and Symonds (2006) Handb Exp Pharmacol. 173, 289-303G; targeting of deoxyribonucleotide sequences is described in Helene, C., et al. (1992) Ann. NY Acad. Sci. 660, 27-36; Maher (1992) Bioassays 14 (12): 807-15; aptamers are described in Lee et al. (2006) Curr Opin Chem Biol. 10, 1-8; RNAi is described in Reynolds et al. (2004) Nature Biotechnology 22 (3), 326-330; Pushparaj and Melendez (2006) Clinical and Experimental Pharmacology and Physiology 33 (5-6), 504-510, which describes RNAi; Dillon et al. (2005) Annual Review of Physiology 67, 147-173, which describes RNAi; Dykxhoorn and Lieberman (2005) Annual Review of Medicine 56, 401-423, which describes RNAi). RNAi molecules are commercially available from various sources (e.g., Ambion, TX; Sigma-Aldrich, MO; Invitrogen).Some siRNA molecular design programs using different algorithms are known in the art (see, for example, Cenix algorithm, Ambion; BLOCK-iT™ RNAi Designer, Invitrogen; siRNA Whitehead Institute Design Tools, Bioinformatics & Research Computing).The defining characteristics of optimal siRNA sequence include the G / C content at the end of siRNA, the Tm of specific internal domain of siRNA, the length of siRNA, the position of target sequence in CDS (coding region) and the nucleotide content of 3' overhang.

[0038] The definitions and methods set forth herein are presented to better define the present disclosure and for guidance to those of ordinary skill in the art in the practice of the present disclosure. Unless otherwise specified, terms are to be understood according to conventional usage by those of ordinary skill in the art.

[0039] In some embodiments, numbers expressing quantities of ingredients, properties such as molecular weight, reaction conditions, and the like used to describe and claim certain embodiments of the present disclosure are understood to be modified in some instances by the term "about." In some embodiments, the term "about" is used to indicate that a value includes the standard deviation of the average for the device or method being employed to determine that value. In some embodiments, the numerical parameters set forth in the specification and appended claims are approximations that may vary depending on the desired properties sought to be obtained by a particular embodiment. In some embodiments, the numerical parameters should be construed in light of the number of reported significant digits and by applying ordinary rounding techniques. Notwithstanding that the numerical ranges and parameters setting forth the broad scope of some embodiments of the present disclosure are approximations, the numerical values ​​set forth in the specific examples are reported as precisely as practicable. The numerical values ​​presented in some embodiments of the present disclosure may contain certain errors necessarily resulting from the standard deviation found in their respective testing measurements. The recitation of ranges of values ​​herein merely serves as a shorthand method of referring individually to each separate value falling within the range. Unless otherwise specified herein, each individual value is incorporated herein as if individually set forth herein.

[0040] In some embodiments, the terms "a," "an," and "the," and similar references, as used in the context of describing particular embodiments (particularly in certain contexts of the claims below), may be construed to encompass both the singular and the plural, unless otherwise indicated. In some embodiments, the term "or," as used herein, including in the claims, is used to mean "and / or," unless it is expressly stated that only alternatives are referred to or that alternatives are mutually exclusive.

[0041] The terms "comprise," "have," and "include" are open-ended linking verbs. Any form or tense of one or more of these verbs, such as "comprises," "comprising," "has," "having," "includes," and "including," are also open-ended. For example, any method that "comprises," "has," or "includes" one or more steps is not limited to having only those one or more steps and may include other unrecited steps. Similarly, any composition or device that "comprises," "has," or "includes" one or more features is not limited to having only those one or more features and may include other unrecited features.

[0042] All methods described herein can be performed in any suitable order unless otherwise specified herein or clearly contradicted by context. Any examples provided herein with respect to a particular embodiment, or the use of illustrative language (e.g., "for example / such as"), are intended solely to better illuminate the disclosure and do not pose a limitation on the scope of the disclosure as otherwise claimed. No language in the specification should be construed as indicating any non-claimed element essential to the practice of the disclosure.

[0043] Grouping of alternative elements or embodiments of the present disclosure disclosed herein is not to be construed as limiting. Each group member may be referred to and claimed individually or in any combination with other members of the group or other elements found herein. One or more members of a group may be included in or excluded from a group for reasons of convenience or patentability. When any such inclusion or exclusion is made, the specification herein includes the modified group and is therefore deemed to satisfy all Markush group descriptions used in the appended claims.

[0044] As will be understood based on the foregoing specification, the above-described aspects of the present disclosure can be implemented using computer programming or operating techniques, including computer software, firmware, hardware, or any combination or subset thereof. Any such resulting program having computer-readable code means can be embodied or provided in one or more computer-readable media, in accordance with the discussed aspects of the present disclosure, to thereby create a computer program product, i.e., an article of manufacture. The computer-readable medium may be, for example, but not limited to, a fixed (hard) drive, a diskette, an optical disk, a magnetic tape, a semiconductor memory such as a read-only memory (ROM), and / or any transmission / reception medium, such as the Internet or other communications network or link. An article of manufacture containing the computer code can be created and / or used by executing the code directly from one medium, by copying the code from one medium to another, or by transmitting the code over a network.

[0045] These computer programs (also known as programs, software, software applications, "apps," or code) contain machine instructions for a programmable processor and may be implemented in a high-level procedural and / or object-oriented programming language and / or in assembly / machine language. As used herein, the terms "machine-readable medium" and / or "computer-readable medium" refer to any computer program product, apparatus, and / or device (e.g., magnetic disk, optical disk, memory, programmable logic device (PLD)) used to provide machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. However, "machine-readable medium" and "computer-readable medium" do not encompass transitory signals. The term "machine-readable signal" refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0046] As used herein, a processor may include any programmable system, including systems that use microcontrollers, reduced instruction set circuits (RISC), application specific integrated circuits (ASIC), logic circuits, and any other circuit or processor capable of performing the functions described herein. The above examples are merely examples, and thus are not intended to limit in any way the definition and / or meaning of the term "processor."

[0047] As used herein, the terms "software" and "firmware" are used interchangeably and encompass any computer program stored in memory, including RAM memory, ROM memory, EPROM memory, EEPROM memory, and non-volatile RAM (NVRAM) memory, for execution by a processor. The above memory types are merely examples and are thus not limiting as to the types of memory usable for storing computer programs.

[0048] In one aspect, a computer program is provided, the program embodied on a computer-readable medium. In one aspect, the system runs on a single computer system, and no connection to a server computer is required. In a further aspect, the system runs in a Windows environment (Windows is a registered trademark of Microsoft Corporation, Redmond, Washington). In yet another aspect, the system runs in a mainframe environment and a UNIX server environment (UNIX is a registered trademark of X / Open Company Limited, Reading, Berkshire, United Kingdom). The application is adaptable and designed to run in a variety of different environments without compromising any primary functionality.

[0049] In some embodiments, a system includes multiple components distributed across multiple computing devices. One or more components may be in the form of computer-executable instructions embodied on a computer-readable medium. The systems and processes are not limited to the specific embodiments described herein. Furthermore, each system and process component can be implemented separately, independent of the other components and processes described herein. Each component and process can also be used in combination with other assembly packages and processes. This embodiment can enhance the functionality and capabilities of a computer and / or computer system.

[0050] All publications, patents, patent applications, and other references cited in this application are herein incorporated by reference in their entirety for all purposes to the same extent as if each individual publication, patent, patent application, or other reference was specifically and individually indicated to be incorporated by reference in its entirety for all purposes. The citation of a reference herein should not be construed as an admission that such reference is prior art to the present disclosure.

[0051] Having described the present disclosure in detail, it will be apparent that modifications, variations, and equivalent embodiments are possible without departing from the scope of the present disclosure as defined in the appended claims. Furthermore, it should be understood that all examples in this disclosure are presented by way of non-limiting examples. [Example]

[0052] The following non-limiting examples are presented to further illustrate the present disclosure. It should be understood by those skilled in the art that the techniques disclosed in the examples follow representative approaches found by the inventors to function well in the practice of the present disclosure and, therefore, can be considered to constitute exemplary modes for practicing the same. However, those skilled in the art should, in light of the present disclosure, understand that many changes can be made in the specific embodiments disclosed and still obtain a like or similar result without departing from the spirit and scope of the present disclosure.

[0053] Example 1 T cell characteristics associated with toxicity to immune checkpoint blockade in melanoma patients The following example describes a method for predicting the likelihood of severe immune-related adverse events (irAEs) occurring in melanoma patients following immunotherapy.

[0054] summary: Severe immune-related adverse events (irAEs) occur in up to 60% of melanoma patients treated with immune checkpoint inhibitors (ICIs). However, it is unclear whether a common baseline immunological condition precedes the onset of irAEs. In this example, we applied time-of-flight mass cytometry, single-cell RNA sequencing, single-cell V(D)J sequencing, bulk RNA sequencing, and bulk T cell receptor (TCR) sequencing to examine peripheral blood samples from melanoma patients treated with anti-PD-1 monotherapy or anti-PD-1 and anti-CTLA-4 combination ICIs. By analyzing 93 blood samples from patients before and during the early stages of ICI treatment, as well as three patient cohorts (n = 27, 26, and 18), we found that two pretreatment factors in the circulation—the abundance of activated CD4+ memory T cells and TCR diversity—were associated with the occurrence of severe irAEs regardless of organ system involvement. We also explored changes in TCR clonality during treatment among patients receiving combination therapy, and correlated findings with the severity and timing of irAE onset. These results demonstrate a circulating T cell signature associated with ICI-induced toxicity and suggest improvements in diagnostic and clinical management.

[0055] result: This study systematically evaluated peripheral blood immunological features associated with ICI-induced toxicity in patients with metastatic melanoma. Across distinct single-cell and bulk profiling modalities, common T cell features associated with the development of severe irAEs within 3 months of treatment initiation were identified. These features were independent of important clinical variables, including durable clinical responses and treatment with anti-PD-1 monotherapy or anti-PD-1 and anti-CTLA-4 combination therapy. Leveraging these findings, we developed predictive models for irAE development and explored their utility for identifying ICI-induced toxicity early before and during treatment.

[0056] Clinical cohort characteristics. To test for candidate risk factors associated with the occurrence of severe (grade 3+) irAEs, 78 patients with metastatic melanoma were identified, of which 71 were evaluable after applying exclusion criteria (Figure 1). Of these patients, 33 were treated with anti-PD-1 monotherapy and 38 with anti-PD-1 plus anti-CTLA-4 combination therapy. Ninety percent had no prior ICI history. All patients were closely monitored for irAE occurrence during and after ICI treatment (median follow-up: 14.9 months; median time to grade 3+ irAE: 1.5 months). The majority of patients experienced one or more irAEs ranging from mild (grade 1) to life-threatening (grade 4), affecting multiple organ systems, which were classified by certified clinicians according to standardized criteria (CTCAE v.5.0). The 71 patients were stratified into three non-overlapping cohorts, a single-cell discovery cohort, and a larger bulk cohort divided into training and validation sets (Figure 1).

[0057] Determinants of severe irAEs from pretreatment blood. Pretreatment peripheral blood was analyzed using mass cytometry to identify intracellular determinants of severe irAEs.

[0058] First, we performed high-dimensional single-cell profiling of pretreatment peripheral blood samples from 18 patients (single-cell discovery cohort, Figures 1 and 2A), eight of whom developed severe irAEs after treatment initiation. By applying time-of-flight mass cytometry (CyTOF) to profile 35 leukocyte markers in each sample, we analyzed 20 distinct subpopulations from nearly 800,000 evaluable cells, encompassing seven major mononuclear lineages: B cells, plasmablasts, CD4 and CD8 T cells, natural killer (NK) cells, natural killer T (NKT) cells, and monocytes (Figures 2B, 2C, 7, and 8A). We then examined each subpopulation with respect to severe irAE outcomes (Figure 2c). Of all subpopulations, only CD4 effector memory T (TEM) cells were significant after multiple hypothesis correction, with higher levels in pretreatment blood associated with severe irAE occurrence (p = 0.0002; Q = 0.004; Figures 2C, 2D, 7B, 8B, and 8C). To confirm this finding, the same peripheral blood samples from 13 patients were investigated using 5' droplet-based 10x chromium single-cell RNA sequencing (scRNA-seq) paired with single-cell V(D)J sequencing (scV(D)J-seq) of TCR and B cell receptor (BCR) clonotypes. After quality control (Figure 7A), the 5' assay yielded 24,807 cells and seven major lineages classified based on reference marker gene expression (Figure 3A). Using unsupervised clustering, we identified 32 distinct transcriptional states across the seven cell types (Figure 3A). We then calculated the association between cell state abundance and the occurrence of severe irAEs. Surprisingly, across these 32 cell states, CD4 T cell state 5, which lacks expression of CCR7 and SELL (CD62L) and is consistent with CD4 TEM cells, was found to be most strongly associated with severe irAEs (nominal P = 0.05, two-tailed, unpaired Wilcoxon rank-sum test; Figure 3B). This state also correlated most strongly with CD4 TEM levels measured by CyTOF (Figure 3B). The joint probability of this result was examined using a permutation test, yielding an empirical P value of 0.003.Further analysis revealed that CD4 T cell state 3, which correlates closely with state 5 by unsupervised hierarchical clustering (Figure 7B), also exhibited an expression profile consistent with CD4 TEM (Figures 3C and 7C). When combined with state 5, the resulting cluster (CD4 T 5+3) was more significantly associated with severe irAE occurrence and CD4 TEM levels enumerated by CyTOF (Figure 3B). Indeed, across all 82 possible pairwise combinations of cell states within each major cell type, CD4 T 5+3 achieved both the highest Spearman correlation with CD4 TEM levels enumerated by CyTOF and the strongest association with severe irAE occurrence (Figures 7D and 7E).

[0059] Differential gene expression analysis relative to other CD4 T cell states revealed that CD4 T 5+3 populations were enriched for markers of activated effector cells, including HLA-DRA, MKI67, TNFRSF4 (OX40), CCL5, and IL32, and depleted for markers of TCM cells (SELL / CD62L) and naive T cells (CCR7, TCF7) (Figures 3C and 7C). Using Seurat Azimuth for reference-derived cell labeling, we confirmed that CD4 T cells were most significantly associated with severe irAEs, similar to the CD4 T 5+3 population identified by de novo analysis (Figure 8). Furthermore, when the CD4 T 5+3 population was subdivided into activated and resting subsets based on the expression of reference activation markers (HLA-DX, MKI67), the activated subset showed the strongest association with the occurrence of severe irAEs (P = 0.002, two-tailed, unpaired Wilcoxon rank-sum test; Figures 3D and 9A). This finding was validated using reference-guided annotation with Azimuth and CyTOF (Figures 9A and 9B), suggesting that activated CD4 T cells preferentially underlie severe ICI toxicity.

[0060] Given this observation, we wondered whether pretreatment TCR diversity in activated CD4 T cells might also correlate with severe ICI toxicity. Indeed, single-cell TCR clonotype diversity (Shannon entropy) of activated CD4 T cells was elevated in patients who developed severe irAEs (area under the receiver operating characteristic curve (AUC) = 0.90, P = 0.05; Figure 3E). This suggests that when activated CD4 T cells are quantified relative to total peripheral blood mononuclear cells (PBMCs), TCR enrichment, defined by the number of unique clonotypes in a sample and key components of diversity metrics, including Shannon entropy, masks loss of diversity due to clonal expansion (Figures 10A and B). In other words, the TCR enrichment of activated CD4 T cells within total PBMCs underlies the overall increase in pretreatment TCR diversity in patients who will develop severe irAEs. Notably, there is substantial precedent in the previous literature for definitions of clonotypic diversity that incorporate abundance, including studies of circulating and tumor-infiltrating T cells, providing a strong basis for their application in the present study.

[0061] When all evaluable T cells were combined, a significant trend was observed between bulk TCR diversity in pretreatment samples and severe irAE occurrence (AUC = 0.80; Figure 3E), although this association was attenuated or absent in other T cell subpopulations. Furthermore, this association was primarily attributable to CD4 T cells with an effector memory profile (low CCR7 and SELL) (Figures 3F and 10C-F). In contrast, the differences in peripheral blood BCR diversity associated with severe irAE occurrence were less pronounced (Figure 10G). Collectively, these findings suggest that a more diverse TCR repertoire in baseline CD4 T cells, broadly reflected in bulk peripheral blood, is associated with the occurrence of severe ICI toxicity.

[0062] Extensive analysis of T cell traits associated with irAEs. Having identified potential pretreatment determinants of severe irAE occurrence, we validated our findings in a larger, independent cohort of patients. Based on sample size estimates, we applied bulk RNA sequencing (bulk RNA-seq) to pretreatment peripheral blood samples from 53 additional metastatic melanoma patients across two cohorts (n = 26 and 27) treated with checkpoint blockade with single agents (anti-PD-1, n = 29) or drug combinations (anti-PD-1 and anti-CTLA-4, n = 24) (Figure 1). To assess circulating immunological signatures in bulk transcriptome profiles, we applied CIBERSORTx, a machine-learning method for enumerating cell subsets from bulk tissue expression profiles, and MiXCR, a computational method for V(D)J clonotype assembly and quantification from bulk RNA-seq data. We confirmed the accuracy of CIBERSORTx for deconvolving major blood lineages, including the specificity of activated CD4 memory T cell (TM) signatures for activated CD4 TEM cells (CyTOF) using peripheral blood from 17 melanoma patients by direct comparison with cytometry assays (Figure 11). Strikingly, among the 13 PBMC subsets evaluable by CIBERSORTx, only activated CD4 TM cell levels were associated with severe irAE occurrence (Figure 4A; P<0.025, HR=8.3 and 14.8 for combination and PD-1, respectively; Figure 14B and C) or across cohorts by LOOCV (P=0.0028 and HR=12.2 for combination therapy; Figure 5A; P=0.03 and HR=9.0 for PD-1 treatment). This model also independently predicted time to severe irAE in a multivariate model of treatment type, age, sex, and other important parameters.

[0063] Peripheral TCR clonal expansion associated with severe irAEs. Previous case reports of melanoma patients who experienced fatal ICI-mediated toxicity have shown evidence of clonally expanded autoreactive or virus-reactive T cells in affected tissues, linking self- and pathogen-recognizing T cell clones to lethal toxicity. Therefore, we hypothesized that pretreatment TCR clonotypes in peripheral blood may be more prone to expansion in patients who will develop severe irAEs after initiating ICI treatment. To investigate this, we applied immunoSEQ to profile bulk TCR-β repertoires in paired pretreatment and early-stage PBMC samples collected from 15 metastatic melanoma patients treated with combination therapy. Using a TCR clonality index (Pielou's equilibrium), which is robust to variations in the number of captured clones, we confirmed significant concordance between MiXCR (bulk RNA-seq) and immunoSEQ (DNA) for pretreatment samples from these 15 patients, thereby highlighting the integrity of the combined model in bulk cohorts 1 and 2 (Figure 15A). We then assessed TCR clonal expansion (i.e., clonal dominance) after treatment initiation, as measured by an increase in 1-Pielou's equilibrium. Supporting our hypothesis, we observed both significantly increased TCR clonal expansion and persistence of baseline clones in patients who developed severe irAEs compared with patients who did not develop severe irAEs (Figures 5B, 15B, and 15C). In patients with severe irAEs (n = 3) who underwent scRNA-seq and single-cell TCR sequencing (scTCR-seq), we observed preferential expansion of the activated CD4 T cell compartment among clones detected in both blood draws (Figures 15D–G). Furthermore, persistent CD4 T cell clones were highly enriched for the CD4 T 5+3 population, as identified by scRNA-seq analysis ( Figures 5C and 15E ).

[0064] We further explored whether the extent of TCR clonal expansion early during treatment correlated with the timing of severe irAE occurrence. Indeed, severe irAEs occurred earlier in patients with greater TCR clonal expansion, whether assessed by tertiles using the log-rank test or by rank using Cox proportional hazards regression (P = 0.003, log-rank test; Figure 5D). These results were significantly independent of the time between blood draws when the analysis was restricted to on-treatment blood draws obtained within 1 month of ICI cycle 1 (Figure 5H).

[0065] Circulating leukocytes in autoimmune diseases. Finally, we asked whether the baseline peripheral blood profiles of patients at risk for developing severe irAEs correspond to clinical autoimmunity. To this end, we applied CIBERSORTx to examine 15 leukocyte subsets in the bulk peripheral blood transcriptome across six studies and 587 patients with either systemic lupus erythematosus (SLE) or inflammatory bowel disease (IBD) versus 191 healthy controls. Using a meta-analysis framework to incorporate P values ​​across studies and pathological deviations (Figure 16), we found that circulating activated CD4 TM cells were most significantly associated with autoimmune disorders compared with healthy individuals (Figure 6). These data suggest that severe irAEs may represent a subclinical or latent autoimmune state that manifests clinically with ICI administration, consistent with recent case reports and multicenter data showing that autoimmune patients treated with immune checkpoint blockade are prone to developing flares as an autoimmune manifestation.

[0066] Consideration This study identified two baseline features—the abundance of activated CD4 TM cells in peripheral blood and a clonally more diverse TCR repertoire—as promising determinants of ICI-induced irAEs in patients with metastatic melanoma. Previous studies have linked (1) activated T cells and clonally expanded TCRs in postmortem tissue to fatal irAEs (myocarditis, encephalitis) and (2) effector CD4 T cells to organ-specific irAEs (destructive thyroiditis, hepatitis). This study extends these findings to pretreatment T cell characteristics in irAE development in various organ systems. Incorporating these features into a combined model predicted a higher risk of severe irAEs and demonstrated sufficient granularity to distinguish between different irAE grades and burdens.

[0067] A significant correlation was also identified between early T-cell clonal expansion and the timing of severe irAE onset in patients treated with the combination therapy. Future studies are needed to further characterize this finding and elucidate the relative contributions of CD4 and CD8 T cells to irAE-associated clonal dynamics.

[0068] Consistent with the possibility of a common immunological mechanism underlying both irAE development and autoimmunity, we also observed elevated levels of activated CD4 TM cells in patients with SLE or IBD. Although it is reasonable to expect that patients with preexisting autoimmunity would have higher activated CD4 T cell levels and a higher rate of severe ICI-induced irAEs, none of the patients in this cohort had documented pre-existing autoimmunity. Furthermore, compensatory immunoregulatory mechanisms that alter baseline irAE risk may develop in such patients before ICI initiation. Nevertheless, it will be important to further examine this link in future studies to determine whether circulating activated CD4 TM cells exhibit an increased propensity to recognize autoantigens in patients at risk for severe ICI toxicity. Indeed, the risk of flares is higher in patients with autoimmune diseases treated with combination immunotherapy, particularly those with gastrointestinal or rheumatic conditions. More reliably identifying these at-risk patients during ICI decision-making may improve their outcomes.

[0069] This study has several limitations. First, it used a retrospective design using archived clinical samples. Second, patients received either anti-PD-1 monotherapy or anti-PD-1 plus anti-CTLA-4 combination therapy, which are associated with different risk profiles for the development of severe irAEs. Third, the majority of irAEs occurred within the first 3 months of initiating ICI treatment, although some occurred later. The median time to the onset of severe irAEs in our cohort was 6.4 weeks (consistent with clinical trial data), and no irAEs occurred after 3 months. Therefore, it remains to be investigated whether this finding generalizes to late-onset irAEs. Fourth, the timing of intratreatment peripheral blood sampling relative to the initiation of treatment during immunotherapy was not uniform. Finally, it remains unclear whether the findings generalize to ICI-associated irAE risk in other cancer types.

[0070] Future studies should address these limitations with greater application of single-cell profiling, both before and during immunotherapy. Furthermore, it will be important to confirm our findings in larger, multicenter cohorts and assess whether circulating immunological determinants of ICI-induced toxicity vary depending on the organ most likely to be involved. If prospectively validated, these findings could facilitate treatment adaptation and improve the risk profile of immune checkpoint blockade, with implications for the prediction and potential prevention of ICI-mediated toxicity.

[0071] method Study design and participants. The samples analyzed in this study were collected with informed consent for research use in accordance with the Declaration of Helsinki (2013) as part of an observational registry focused on melanoma, and were approved by the Institutional Review Boards of Yale University School of Medicine and Washington University School of Medicine. Eligible patients aged 18 years or older with metastatic melanoma were treated with ICI therapy consisting of either anti-PD-1 blockade (nivolumab or pembrolizumab) or a combination of immune checkpoint blockade (anti-PD-1 (nivolumab) and anti-CTLA-4 (ipilimumab); Figure 1). Ninety percent of patients had not previously received any immune checkpoint blockade at the time of pretreatment blood collection. All patients underwent routine clinical assessment by a board-certified medical oncologist for irAEs and response. Surveillance was performed before each cycle of ICI treatment (approximately every 3 weeks) and in some cases more frequently (e.g., by inpatient care providers for patients hospitalized with severe irAEs). Surveillance also continued after completion of the treatment course, if applicable. All irAEs were classified according to the Common Terminology Criteria for Adverse Events (CTCAE) v.5.0 from the United States Health and Human Services, with grades ≥2 and ≥3 considered symptomatic and severe, respectively. Within and across patient cohorts, irAEs spanned multiple organ systems, including the gastrointestinal tract, skin, liver, pituitary, thyroid, adrenal, musculoskeletal, ocular, pancreatic, and cardiac systems (Figure 13I). Three patients developed systemic inflammatory syndromes associated with ICI administration (YUGIM, YUHERN, and YUTORY). All severe irAEs occurred within 3 months of ICI initiation, a landmark period during which no patients in this cohort experienced death. Responses were scored as previously defined: durable clinical benefit, no durable benefit, or inevaluable.Three patient cohorts meeting the above-mentioned eligibility criteria were identified, and pretreatment PBMC samples were collected immediately prior to the first cycle of anti-PD-1 or combination ICI administration (median day 0; range 0-2 months). PBMCs from each cohort (pretreatment for all patients and pretreatment / on-treatment paired for 15 patients) were analyzed, as shown in Figure 1.

[0072] Blood collection and processing. Peripheral blood samples were collected in K2EDTA Vacutainer tubes (Becton Dickinson) and processed within 1 hour of collection. PBMC extraction was performed using either the ammonium chloride protocol or the Lymphoprep (STEMCELL Technologies) protocol. The Lymphoprep protocol was applied according to the manufacturer's instructions. When using the ammonium chloride protocol, 4–8 ml of blood was mixed with 20 ml of cold ammonium chloride lysis buffer (0.1 M ammonium chloride, 0.01 M Tris-HCl) and incubated at room temperature for 5 minutes. Cells were then centrifuged at 300 g for 5 minutes and washed with 5 ml of cold PBS. PBMC samples were cryopreserved in 10% dimethyl sulfoxide / 90% FBS. Cryovials were placed in Nalgene Mr. Frosty containers (Thermo Fisher Scientific) for 24 hours and then stored in liquid nitrogen until cell and RNA processing for expression analysis.

[0073] Mass cytometry. Metal-conjugated antibodies were purchased pre-conjugated from Fluidigm or purified from BioLegend, Thermo Fisher Scientific, or Cell Signaling Technology and then conjugated to metals using Maxpar Antibody Labeling Kits (Fluidigm) according to the manufacturer's instructions. PBMCs from each of 28 patients were prepared for CyTOF. Frozen cell suspensions were first thawed by placing the cryovials in a 37°C water bath without immersing the caps for 1–2 min. Then, 1–3 × 10 cells were collected in single-cell suspensions. 6 PBMCs were incubated with Human TruStain FcX (BioLegend) for 10 minutes at room temperature to block nonspecific antibody binding, followed by incubation with a metal-conjugated antibody against a cell surface molecule on ice for 20 minutes. Cells were also incubated with Cell-ID Cisplatin (Fluidigm) according to the manufacturer's instructions to identify viable cells. After treatment with intracellular fixation and permeabilization buffers (Thermo Fisher Scientific), cells were incubated with a metal-conjugated antibody against an intracellular protein. Cells were then washed, stained with Cell-ID Intercalator-Ir (Fluidigm) diluted in PBS containing 1.6% paraformaldehyde (Electron Microscopy Sciences), and stored at 4°C until acquisition. After the wash step, sample acquisition was then performed using a Helios System (Fluidigm) for <400 s. -1 The event rate was 1.0.

[0074] To reduce technical variation between samples, Ce beads were used for each sample, and files were normalized collectively using Bead Normalizer v0.3 (https: / / github.com / nolanlab / bead-normalization / wiki / Installing-the-Normalizer). To further minimize technical variability, sample processing and acquisition batches were limited to four, the same reagent lot was used across all samples, and no major adjustments to the Helios calibration were made. We also note that Astrolabe does not compare numerical intensities between samples; instead, each sample was analyzed separately, assuming that a given subset is the same regardless of whether the underlying marker intensities shift. Therefore, this platform is reported to be resistant to batch effects.

[0075] Mass cytometry data analysis. CyTOF data were initially analyzed using Cytobank v8.0 and v8.1 (Beckman Coulter) with the FlowSOM algorithm for hierarchical cluster optimization and the viSNE algorithm (iterations = 5,000, perplexity = 100) for visualization of high-dimensional data. Subsequent identification of cell subpopulations and data visualization were performed using the Astrolabe Cytometry Platform v3.6 and v4.0 (Astrolabe), which utilizes the Ek'Balam algorithm, which combines a knowledge-based hierarchical annotation strategy with unsupervised clustering to automatically label cell subpopulations. In total, 20 cell subpopulations spanning the major mononuclear lineages in peripheral blood were identified and quantified. For each patient sample, cell subpopulation levels were normalized to a sum of 1, and cells that could not be classified based on protein marker expression were excluded from the analysis. To validate Astrolabe, blinded manual gating of major cell populations, including CD4 TEM cells, was performed using Cytobank (Figures 17, 18A, and 18B). Total abundance of CD4 TEM cells, calculated as a fraction of either total PBMCs or circulating T cells, but not as a fraction of CD4 T cells, was significantly associated with severe irAE occurrence (Figure 18C).

[0076] Flow cytometry. PBMCs from five healthy donors were analyzed by flow cytometry (Figure 11E). Briefly, 2–5 million PBMCs were treated with TruStain FcX Fc Receptor Blocking Solution (BioLegend) for 10 min at room temperature to block Fc receptors, and then stained with fluorophore-tagged surface antibodies for 30 min at room temperature. Cells were stained using the following antibodies: anti-human CD45 conjugated with FITC (clone 2D1; BioLegend); anti-human CD3 conjugated with AF700 (clone OKT3; BioLegend); anti-human CD4 conjugated with APC (clone OKT4; BioLegend); anti-human CD8 conjugated with PE / Cy7 (clone SK1; BioLegend); anti-human CD19 conjugated with APC-Cy7 (clone HIB19; BioLegend); anti-human CD14 conjugated with PerCp / Cy5.5 (clone HCD14; BioLegend); and anti-human CD56 conjugated with BV605 (clone 5.1H11; BioLegend). Cells were then washed twice with ice-cold PBS-based buffer (1x PBS, 2% FBS, 1 mM EDTA) and stained with 4',6-diamidino-2-phenylindole (DAPI) (BioLegend) to assess cell viability. Antibody capture beads (BD Biosciences) were used to offset each fluorophore in the experiment. Stained cells were analyzed by flow cytometry with operator assistance using a MoFlo Legacy instrument (Beckman Coulter) at the Siteman Flow Cytometry Core at Washington University School of Medicine. After excluding DAPI-positive cells and putative doublets based on forward and side scatter analysis, major lymphocyte populations, including B cells, CD4 T cells, CD8 T cells, and NK cells, were enumerated as percentages of total lymphocytes using FlowJo v.10 (FlowJo LLC).

[0077] scRNA-seq and scV(D)J-seq library preparation and sequencing. Single-cell suspensions from PBMC samples were obtained as described above and counted using a hemocytometer (Thermo Fisher Scientific) or a Coulter Counter (Beckman Coulter Life Sciences) according to the manufacturer's instructions at a concentration of 700–1,200 viable cells per μl. The single-cell suspensions were then subjected to library preparation using a 5' transcriptome kit (10x Genomics) according to the manufacturer's instructions for scRNA-seq paired with scV(D)J-seq for TCR and BCR clonotypes. Complementary DNA libraries were sequenced on a NovaSeq instrument (Illumina) with 2 × 92 base pair (bp) paired-end reads targeting an average of 20,000 reads per cell.

[0078] scRNA-seq analysis (discovery cohort). Raw scRNA-seq reads were de-duplicated by barcode and aligned to the hg38 reference genome using Cell Ranger v.3.1.0 to generate low-density digital count matrices, which were then analyzed using Seurat v.3.1.5 or v.3.2.1 (72) to identify cell types and cell states. Outlier cells were identified and excluded based on the following criteria: (1) mitochondrial content >25% or (2) cells with fewer than 100 or more than 1,500–3,000 expressed genes, depending on the sample-level distribution. After normalization (NormalizeData) and variable feature identification (FindVariableFeatures, n = 2,000 features), anchors were identified using FindIntegrationAnchors (dims = 1:30), and batch correction was performed using IntegrateData (with default parameters). Once integrated, we applied principal component analysis (PCA) and uniform manifold approximation projection (UMAP) using the 2,000 most variable genes and the top 30 principal components. We applied FindClusters with a resolution parameter set to 3 to identify cell types and cell states, resulting in 37 clusters.

[0079] All identified clusters were assigned to major cell lineages based on the expression of reference marker genes: CD3D / CD3Ehi = T cells; CD8A / CD8Bhi and NKG7 / GNLYlo = CD8 T cells; non-CD8 T cells with high IL7R expression and low NKG7 / GNLY = CD4 T cells; NKG7 / GNLYhi and CD3D / CD3Elo = NK cells; CD14 or FCGR3Ahi = monocytes; FCER1Ahi = dendritic cells (DCs); MS4A1hi = B cells; HBBhi = red blood cells; and PPBPhi = platelets. Cells with high expression of CD3D / E and GNLY / NKG7 that did not annotate CD8 / CD4 T cells were included in the T or NKT cell group and designated T / NKT. Clusters annotated as red blood cells or platelets were excluded from further analysis. To assess the effective doublet rate, cell barcodes were cross-referenced with single-cell BCR (scBCR) and TCR (scTCR) clonotypes. By determining (1) the percentage of non-T cells that mapped abnormally to a TCR clonotype (denoted m) and (2) the frequency (i.e., recovery rate) of T cells annotated with the matching scTCR clonotype (denoted f), we calculated the effective doublet rate (m / f) to be 2.2%. The calculated effective doublet rate for scBCR clonotypes mapped to non-B cells was the same (also 2.2%). Because the effective doublet rate was reasonably low, we excluded all single cells with aberrant expression of TCR or BCR clonotype sequences. PCA, UMAP, and FindClusters were then repeated as described above, yielding 32 clusters. Two erythrocyte clusters characterized by very high HBB expression were retained and excluded from the analysis. Subsequently, one final round of PCA, UMAP, and FindClusters was performed, yielding a final set of 32 clusters (i.e., states). The low-dimensional embedding is shown in Figures 3A and 7A.

[0080] All 32 conditions were assessed for their association with severe irAE occurrence (x-axis in Figure 3B) and CD4 T cell abundance measured by CyTOF (y-axis in Figure 3B). Among them, CD4 T cluster 5 correlated most strongly with both variables (Figure 3B). To determine the statistical significance of this result, we calculated the joint probability of (1) being ranked first by each criterion and (2) achieving a P value and Spearman correlation coefficient at least as strong as that of CD4 T cluster 5. To empirically calculate this probability, we implemented a permutation scheme that shuffled the cell fractions associated with each scRNA-seq cluster independently across all patient samples and then evaluated (1) and (2) above. Repeating this process 10,000 times yielded an empirical P value of 0.003 for CD4 T cluster 5. Pairwise combinatorial analysis was also performed, restricting cell state pairs to the same major cell type (B cells, CD4 T cells, CD8 T cells, NK cells, and monocytes) to maintain biological consistency. Each of the 82 possible cell cluster combinations was compared with CD4 T cell TEM levels and severe irAE occurrence enumerated by CyTOF (Figures 7D and 7E). CD4 T cell clusters 5 and 3 emerged as the top pair. Using the statistical methods described above, an empirical P value of 0.002 was calculated for this result. To identify differentially expressed genes (DEGs) in Figure 3C, Seurat FindMarkers was applied with default parameters to the CD4 T 5+3 population and the other CD4 T cell states.

[0081] To assess the relative utility of unsupervised clustering for delineating the intracellular determinants of irAE development, we leveraged the reference-guided annotation framework within Seurat v.4.0.1 (Azimuth) to project our scRNA-seq dataset onto a PBMC atlas of 161,764 cells across six major lineages and 27 finer-grained subsets defined using scRNA-seq and co-detection of over 220 protein markers. First, we preprocessed the query dataset according to the quality control steps described above, yielding 24,807 cells. Then, we normalized the query dataset using SCTransform, and applied FindTransferAnchors to the query and reference datasets using a 50-dimensional pre-computed supervised PCA transform. Then, we applied MapQuery to map cell type labeling and UMAP structure by referencing the query dataset.

[0082] Among the 27 cell states identified by Azimuth (Figure 8A), CD4 TEM was most strongly associated with severe irAE development and correlated most strongly with CD4 TEM cells enumerated by CyTOF (Figure 8C). Among the two other CD4 TEM-like subsets identified by Azimuth (CD4 CTL, CD4 proliferative), CD4 proliferative showed the highest expression of HLA-DX and the lowest expression of SELL (Figure 8D), consistent with an activated CD4 TEM phenotype. Furthermore, when protein expression was examined by Azimuth using antibody-derived tag data, only the CD4 TEM and CD4 proliferative states exhibited TEM cell characteristics (CD45ROhiCD45RAloCD27lo; Figure 8E). Indeed, the combined CD4 TEM and CD4 proliferative populations were most strongly associated with severe irAE development (Figure 8C). A hypergeometric test was applied to assess the overlap of cell barcodes between the combined CD4 T+ CD4 proliferative population (Azimuth) and states defined by de novo clustering. CD4 T5+3 emerged as the top hit (Benjamini-Hochberg adjusted P = 2.5 × 10-7 Despite strong overlap between unsupervised and supervised approaches, CD4 T 5+3 were associated with greater and more severe irAE incidence and CyTOF than populations labeled by reference-guided annotation (Figure 8F).

[0083] Bulk RNA-seq library preparation, sequencing, and quantification. Cryopreserved cell suspensions were thawed as described above. RNA was then extracted using the RNeasy PowerLyzer Tissue & Cells Kit (QIAGEN) and quality assessed using the 2100 Bioanalyzer System (Agilent Technologies). All samples were of sufficiently high quality for TruSeq RNA Exome analysis (DV200 > 30%) and were prepared using the TruSeq RNA Exome Kit (Illumina) according to the manufacturer's instructions. After hybrid capture, cDNA libraries were pooled and sequenced on a HiSeq 2500 instrument (Illumina) using 2 × 150 bp paired-end reads, targeting 20–25 million reads per sample. Raw reads were quantified using Salmon v.0.12.0 with the GENCODE v.29 reference transcriptome; the following command line arguments were used, with the rest using default parameters: --seqBias --gcBias --posBias --validateMappings --rangeFactorizationBins 4. Read counts were normalized to transcripts per million (TPM) at the gene level using tximport v.1.10.1. Only samples with a mapping rate of ≥60% and successful TCR assembly (see V(D)J receptor profiling and clonotype analysis below) were included in further analysis, with the exception of three samples with a mapping rate of >40% (but <60%) and successful TCR assembly. In total, 53 sequenced samples (88%) within bulk cohorts 1 and 2 met these criteria (Figure 1).

[0084] Bulk RNA-seq deconvolution. To determine the leukocyte composition in bulk RNA-seq profiles of PBMCs, CIBERSORTx v.1.0.41 (https: / / cibersortx.stanford.edu) was applied to the TPM matrix of each cohort along with the LM22 signature matrix (Figure 1). CIBERSORTx was applied separately to each sequencing batch, with B-mode batch correction and without quantile normalization. LM22, consisting of highly optimized reference profiles for distinguishing 22 functionally defined human hematopoietic subsets, has been extensively validated for flow cytometry to accurately enumerate leukocyte subsets in whole blood and PBMCs, whether profiled by RNA-seq or microarrays. The performance of CIBERSORTx and LM22-based activated CD4 TM cell profiling was further validated in this study by gene expression analysis (CCR5, SELL, TCF7, and CD27; Figure 11A). Using PBMC samples from melanoma patients, comparisons were made between CIBERSORTx, mass cytometry, flow cytometry, and scRNA-seq (Figures 11B, 11C, 11D, 11E, and 11F). All LM22 subsets except granulocyte and macrophage subsets were evaluated in this study (n = 15; Figure 4A), and their relative fractions in each sample were renormalized to a sum of 1. Although a total of 15 subsets were evaluated, two were detected at low density by CIBERSORTx (regulatory T (Treg) cells, gamma delta T cells) and could not be assessed by the Wilcoxon rank-sum test in Figure 4A.

[0085] V(D)J receptor profiling and clonotype analysis. For the single-cell discovery cohort, raw scV(D)J-seq reads were mapped to the reference refdata-cellranger-vdj-GRCh38-altsensembl-4.0.0 using Cell Ranger v.3.1.0, and the resulting clonotype assemblies were downloaded from Loupe V(D)J browser v.3.0.0 (10× Genomics). Given that activated TM cells arise through clonal expansion, activated TM cells are expected to have lower TCR diversity than their naive counterparts, provided either (1) cells from both populations are equally sampled (i.e., their counts are equivalent) or (2) variation in total T cell counts is eliminated by normalization (Figure 10A). However, by ignoring variation in total T cell frequency, such sampling ignores abundance, which is the number of unique species (clonotypes) within a population and is an important factor underlying immune repertoire diversity. Therefore, in this study, we first used Shannon entropy, a theoretical metric that combines evenness and richness as a single evaluation criterion, to characterize immune repertoire diversity (Figure 10A). For each evaluable patient sample within the single-cell discovery cohort (Figures 1 and 2A), we randomly sampled (non-recovered) the TCR clonotype repertoire to equalize the number of evaluable PBMC cells across patients while also addressing technical variability in TCR recovery rates. To maximize the pool of TCR clones available for sampling, we excluded patients with fewer than 100 TCR clones (n = 4; YUTAUR, YUTORY, YUHERN, and YUTHEA). Then, for the remaining nine patients, we calculated Shannon entropy (R package vegan v.2.5-6 (reference 78)) for each T cell subset relative to total PBMCs, and the resulting values ​​were averaged over 100 iterations of this procedure (Figures 3E, 10B, and 10D-F). For the same nine patients, Shannon entropy was analyzed as above for scBCR clonotypes across the IGK, IGL, and IGH chains (Fig. 10G).

[0086] For bulk cohorts 1 and 2, TCR clonotypes were assembled and quantified using MiXCR v.3.0.125 after trimming adapter sequences using Skewer v.0.2.2, using the following commands: mixcr align -p rna-seq -s hsa -O allowPartialAlignments=true data_R1.fastq.gz data_R2.fastq.gz alignments.vdjca. For each patient sample, TCR clonotype diversity was measured overall for the TCR-α and TCR-β chains using Shannon entropy (R package vegan v.2.5-6), and comparisons between patients were made based on irAE severity (Figures 4B, 4C, 12A, and 12C). The Gini-Simpson index, calculated using the R package immunarch v.0.6.5 (https: / / doi.org / 10.5281 / zenodo.3367200), was further applied to assess bulk TCR diversity in relation to irAE severity (Figures 12B and 12D). Notably, TCR abundance is a key component for calculating both Shannon entropy and Gini-Simpson index.

[0087] Analysis of T cell clonal dynamics from bulk PBMCs. Bulk TCR-β chain profiling was performed on paired pre-treatment and early-stage on-treatment PBMCs from 15 patients treated with concomitant ICIs. No on-treatment peripheral blood samples were collected after the onset of severe irAEs. Genomic DNA was extracted using the DNeasy Blood & Tissue Kit (QIAGEN) and subjected to survey-resolution immunoSEQ (Adaptive Biotechnologies). Data from the productive TCR-β chain rearrangements were exported using the immunoSEQ Analyzer online tool, and the richness and diversity of the TCR-β repertoire were assessed using Pielou's equilibrium, with increased 1-equilibrium associated with increased clonality. Pielou's equilibrium obtained from immunoSEQ profiling was compared with bulk RNA-seq (MiXCR) and found to be concordant (Figure 15A). Appropriate pairing of all pre-treatment and on-treatment samples was also verified by cross-comparison of TCR-β CDR3 sequences. Clonal expansion was inferred by analyzing the difference in clonality, defined as the 1-Pielou equilibrium, between paired on-treatment and pre-treatment time points in each sample (Figs. 5B and 15B). More specifically, to calculate the change in clonality from baseline, pre-treatment clonality was subtracted from on-treatment clonality in a paired manner, thereby normalizing all pre-treatment samples to zero (Fig. 5B, left). In Fig. 15B, the data from Fig. 5B was also analyzed without normalizing on-treatment samples to their paired pre-treatment samples.

[0088] To assess the absence of severe irAEs, the degree of clonal expansion, denoted δ, was divided equally into tertiles using the R package dplyr v.1.0.7 (Figures 5D and 15H). This resulted in the following groups: no clonal expansion, δ < 0, n = 5; intermediate, 0 < δ < 0.009, n = 5; and high clonal expansion, δ > 0.009, n = 5. These thresholds were applied to the full immunoSEQ cohort (n = 15; Figure 5D) and to patients for whom blood samples were obtained at day 1 and < 1 month during treatment with ICI (n = 7; Figure 15H). Furthermore, when expressed in rank space, the degree of clonal expansion was significantly associated with the time to severe irAE occurrence in a Cox regression model, independent of the time between blood draws, the number of proliferative TCR clones detected, and the age and sex of each patient.

[0089] Analysis of persistent T cell clones. Three patients with severe irAEs and variable levels of clonal expansion: YUALOE, YUNANCY, and YUHONEY (Figures 5C and 15D, 15E, 15F, and 15G) underwent paired pretreatment peripheral blood scRNA-seq and scTCR-seq (Figure 5B). Notably, samples from these three patients had not been previously profiled by scRNA-seq or scV(D)J-seq in the single-cell discovery cohort. Sequencing libraries were generated and processed for quality control as described for the single-cell discovery cohort. Mapping was performed using Cell Ranger v.5.0.1.

[0090] To analyze persistent clones, defined as expanding TCR-β CDR3 nucleotide sequences shared between paired pre- and on-treatment blood samples, immunoSEQ data were examined for shared clonotypes with at least two templates in one blood draw (pre- or on-treatment) and at least one template in the other blood draw (average 60% of all shared clones). This allowed us to prioritize expanding or contracting persistent clones. The resulting sequences were cross-referenced with TCR-β CDR3 nucleotide sequences from the pre-treatment scTCR-seq library, which were further cross-referenced with scRNA-seq data and filtered for cells annotated as T cells by Azimuth (applied as described above) (Figure 15D). In total, 1,504 single-cell transcriptomes with paired immunoSEQ clonotype data were identified. For each patient, a significant Spearman correlation was observed between pretreatment single-cell and immunoSEQ TCR clonotype frequencies (ρ > 0.59; ρ 0 and CD8A / B = 0 for CD4 T cells; CD8A or CD8B > 0 and CD4 = 0 for CD8 T cells). Overall, 69% of all cross-referenced clonotypes could be unambiguously labeled by this approach (Figure 15E). For the plot shown in Figure 5C, the average log2 fold change between CD4 T 5 and 3 was calculated for the remaining CD4 T cell clusters within the single-cell discovery cohort (Figure 7B), and the top 20 genes were then selected for subsequent analysis. Enrichment of this gene set was determined using single-sample GSEA (R package escape v.1.0.1) and applied to T cells labeled by Azimuth or for persistent CD4 / CD8 T cells as described above. For the analyses shown in Figures 15F and G, the proliferative frequencies of persistent T cell clones measured by immunoSEQ were grouped into CD4 and CD8 T cells, and differences in proliferative frequencies were expressed by clonotype (Figure 15G) or overall (Figure 15F) and compared to bulk clonal expansion from baseline (Figure 5B).

[0091] An integrated model for predicting irAE occurrence. Activated CD4 TM cell abundance and bulk TCR clonotype diversity were independently associated with severe irAE occurrence (Figures 4A and B). Therefore, integrative modeling was explored as a means to improve performance. Several techniques were evaluated, including nonlinear modeling using random forests; however, logistic regression (glm in R) achieved comparable performance, and generalized linear models were chosen due to their relative simplicity and robustness. Prior to training, each trait was tested for outliers in bulk cohorts 1 and 2 using the ROUT test at a false discovery rate of 10%. Of 88 data points (2 traits × 53 samples), three outliers were detected, all of which were from activated CD4 TM cells in bulk cohort 1. Regardless of the training cohort, all detected outliers consistently came from these three samples. Therefore, for each integrative model, the maximum fraction of activated CD4 TM cell levels among all non-outlier samples within the training cohort, maxF, was determined. maxF was then used as an upper bound for all samples.

[0092] Combined models were trained to predict the occurrence of severe irAEs (grade 3+) in several ways. These included training on bulk cohort 1 and testing on held-out bulk cohort 2 (Figure 4D, left); training on one treatment regimen and testing on the other (Figure 4D, right); and training across bulk cohorts using LOOCV. For LOOCV evaluation, the analysis was repeated n times for all models, where n is the total number of patients. In each iteration, a model was trained on each patient except the i-th patient and evaluated on the i-th held-out patient. To mitigate overfitting when dividing patients into high and low groups by LOOCV, Youden's J statistic was applied to determine the threshold that optimized sensitivity and specificity in each training cohort, and the i-th held-out patient was then assigned based on this threshold.

[0093] The combined model scores were evaluated by receiver operating characteristic (ROC) analysis. The model trained to distinguish severe from non-severe irAEs was used to predict future severe irAE occurrence (Figures 4D and 13A), irAE grade (Figures 4C, 4E, 13B, and 13E), the number of organ systems affected by irAEs (Figures 13H-J), and time to severe irAE occurrence (Figures 5A and 14). The model was also evaluated in different patient subgroups (Figures 4D and 13D) and compared with pathways assessed by bulk RNA-seq and previously published biomarkers (Figure 13C). The combined model was further validated with different irAE grade thresholds (Figure 13F) and tested separately by treatment type to predict irAE occurrence (Figures 4D, 5A, 13A, 13D-F, 8B, and 8C).

[0094] Assessment of circulating leukocyte composition in autoimmune disorders. Peripheral blood gene expression datasets spanning 239 SLE patients, 348 IBD patients, and 191 paired healthy controls profiled by bulk RNA-seq or microarray were downloaded from Gene Expression Omnibus (GEO). RNA-seq data from Hung et al. were downloaded as preprocessed expression matrices and normalized with TPM before analysis. Affymetrix microarray datasets (n = 5) were downloaded as CEL files, normalized with MAS5 (affy v.3.12 in R (Ref. 82)), mapped to Entrez gene identifiers using custom chip definition files specific to each platform (http: / / brainarray.mbni.med.umich.edu / Brainarray / Database / CustomCDF / ), and converted to HUGO gene symbols. One dataset did not have available raw CEL files; instead, preprocessed expression data was obtained from GEO. When multiple probe sets mapped to the same gene symbol, the probe set with the highest average log2 expression across samples was selected for further analysis. In the Palmer et al. dataset, some samples identified as controls were from subjects with Escherichia coli infection, celiac disease, or progression to Crohn's disease; these were excluded from analysis. The most recent sample was selected as the replicate in the Carpintero et al. dataset. For the Peters et al. dataset, only pretreatment blood samples (week 0) from patients with Crohn's disease were further analyzed. CIBERSORTx49 was applied to the bulk RNA-seq dataset of Hung et al. with LM22 as described above, while the microarray dataset was run with either (1) quantile normalization and B-mode batch correction (non-HG-U133 platform) or (2) quantile normalization and no batch correction (HG-U133 platform). Leukocyte subsets were restricted to mononuclear subsets found in peripheral blood (granulocytes and macrophages were removed) and renormalized to sum to 1 for each sample.

[0095] Within each dataset, a two-tailed, unpaired Wilcoxon rank-sum test was applied to assess the levels of each leukocyte subset in peripheral blood between diseased individuals and healthy controls from the same study (Figure 16). The resulting P values ​​were converted to two-tailed z-scores, taking into account the direction of the association. Within a given disease phenotype (SLE or IBD), z-scores were combined across datasets using the Liptak method, weighted by sample size (Figure 16). Finally, SLE-specific and IBD-specific meta z-scores were combined using the Stouffer method (Figure 16), yielding a pan-SLE / IBD meta z-score for each leukocyte subset (Figure 6).

[0096] Candidate toxicity biomarkers from previous literature and pathway analysis. The combined model was evaluated against previously published irAE biomarkers and enriched pathways for the prediction of severe irAEs (Figure 13C). Each candidate biomarker was evaluated separately in bulk cohorts 1 and 2 by determining the AUC via receiver operating characteristic curve (ROC) analysis. The following pretreatment irAE biomarkers, previously measured by protein expression in the literature, were evaluated in this study as surrogates of peripheral blood RNA: ADPGK and LCP1, assessed individually using bivariate linear regression; CD74 and GNAL15 expression; and the CYTOX score, assessed as the geometric mean expression of genes encoding the same 11 cytokines (CSF3, CSF2, CX3CL1, FGF2, IFNA2, IL12A, IL1A, IL1B, IL1RA, IL2, and IL13). Separately, GSEA v.4.1.0, pre-ranked by GSEAPreranked v.7.1.0, was applied to identify pathways most enriched by irAEs in bulk cohorts 1 and 2 from the Molecular Signatures Database v7.4 hallmark pathway collection. As input, transcriptome-wide gene lists ranked by log2 fold change between patients with and without severe irAEs were defined for bulk cohorts 1 and 2. Gene sets with q<0.25 were considered statistically significant. The two gene sets (MYC_TARGETS_V1; OXIDATIVE_PHOSPHORYLATION) most enriched in patients with severe irAEs versus patients without severe irAEs were compared with the combined model in bulk cohorts 1 and 2 (Figure 13C).

[0097] Statistics. All statistical tests were two-sided unless otherwise specified. The Wilcoxon rank-sum test was used to assess statistical differences between two groups. When assessing more than two groups simultaneously, the nonparametric Kruskal-Wallis test was used. The Benjamini-Hochberg method was applied for multiple hypothesis testing unless otherwise specified. A permutation scheme was implemented to assess scRNA-seq cluster correlation using severe irAE incidence and CyTOF CD4 TEM abundance as described above. Fisher's exact test was applied to assess statistical differences between two categorical variables. Receiver operating characteristic (ROC) analysis was performed to assess classification accuracy and quantified by the area under curve (AUC). Statistical significance of the AUC was determined by a two-sided z-test. Youden's J statistic was used to identify the optimal cutpoint after ROC analysis. Linear agreement was determined by Pearson (r) or Spearman (ρ) correlation, and a two-sided t-test was used to assess whether the results were significantly different from zero. Kaplan-Meier and Cox regression analyses were used to assess covariates related to time to severe irAEs. Significance levels and HRs for Kaplan-Meier analyses were determined using two-sided log-rank tests. The combined models and associated analyses in Figures 5A and 14 included patients from bulk cohorts 1 and 2 (Figure 1) and did not experience severe irAEs, with the exception of two patients (YUDIME and YUMEDIC) who experienced early disease progression and subsequently switched treatment before the 3-month mark. These two patients were included in other analyses because they each received 63 days (2.1 months) of immune checkpoint blockade, the duration during which 76% of all severe irAEs occurred in this patient population.

[0098] For Cox regression, results were analyzed based on Wald statistics (z-scores), and significance was assessed by the Wald test. The proportional hazards assumption was confirmed for each covariate included in the Cox regression, followed by analysis by assessing Schoenfeld residuals. Where appropriate, the Liptak and Stouffer methods were used for pooled statistical analysis. Sample size calculations for bulk cohorts 1 and 2 were performed using pwr v.1.3-0 in R86. In the single-cell discovery cohort, the effect size of the association between CD4 TEM cell abundance (CyTOF) and severe irAE occurrence was 1.99 (Figure 2C and D). Bulk cohorts 1 and 2 were designed to meet this effect size requirement at α = 0.05 and 1 - β = 0.8, while emphasizing specificity (number of patients without severe irAEs > number of patients with severe irAEs) in bulk cohort 1 and balance (number of patients without severe irAEs ≈ number of patients with severe irAEs) in bulk cohort 2. All statistical analyses were performed using R v. 3.5.1+ or Prism 8+ (GraphPad Software).

Claims

1. A method for obtaining a composite score as an indicator of the occurrence of severe immune-related adverse events in a subject receiving an immune checkpoint inhibitor, The steps include: measuring (i) the amount of activated CD4 memory T cells and (ii) a value representing T cell receptor diversity in a biological sample obtained from a subject before administration of an immune checkpoint inhibitor; The steps include: (i) and (ii) combining the measured values ​​into the composite score; A method comprising, wherein if the composite score exceeds a predetermined threshold, the composite score indicates that the subject has an increased risk of severe immune-related adverse events.

2. The method according to claim 1, wherein the predetermined threshold is established from a clinical reference standard.

3. The method according to claim 1, wherein the severe immune-related adverse event is graded to Grade 3 or higher according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.

0.

4. The method according to claim 1, wherein the activated CD4 memory T cells are CCR7-SELL-effector memory T cells.

5. The method according to claim 4, wherein the activated CD4 memory T cells express one or more of HLA-DR, MKI67, and CD38.

6. The method according to claim 1, wherein the value representing the T cell receptor diversity is selected from the group consisting of Shannon entropy, Pierou equilibrium, and the Gini-Simpson index.

7. The method according to claim 1, wherein the amount of activated CD4 memory T cells and the value representing the T cell receptor diversity are measured using at least one of bulk RNA sequencing, time-of-flight mass cytometry, TCR-β profiling, droplet-based scRNA sequencing and scTCR sequencing, and targeted RNA sequencing using an RNA panel targeting activated CD4 memory T cells.

8. A method for obtaining the degree of TCR expansion as an indicator of the occurrence of severe immune-related adverse events in a subject receiving an immune checkpoint inhibitor, The steps include: quantifying the first TCR diversity level derived from a first biological sample obtained from the subject before treatment with immunotherapy; The steps include: quantifying the second TCR diversity level derived from a second biological sample obtained after the administration of the immunotherapy; A step of obtaining the degree of TCR expansion by subtracting the first TCR diversity level from the second TCR diversity level. A method comprising, wherein if the degree of TCR expansion exceeds a threshold, the degree of TCR expansion indicates that the subject is likely to develop a severe immune-related adverse event.

9. The method according to claim 8, wherein the second biological sample is obtained within one month after administration of the immune checkpoint inhibitor.

10. The method according to claim 8, wherein a predetermined threshold is established from a clinical reference standard.

11. The method according to claim 8, wherein the severe immune-related adverse event is graded to Grade 3 or higher according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.

0.

12. The method according to claim 8, further comprising the step of quantifying the amount of activated CD4 memory T cells.

13. The method according to claim 8, wherein the quantification step uses at least one of bulk RNA sequencing, time-of-flight mass cytometry, TCR-β profiling, droplet-based scRNA sequencing and scTCR sequencing, and targeted RNA sequencing using an RNA panel targeting activated CD4 memory T cells.

Images